Project Report PR-PK-87d-02

Pakistan

Social audit of governance and delivery of public services

Baseline Report

Anne Cockcroft, Neil Andersson, Khalid Omer, Noor Ansari, Amir Khan and Ubaid Ullah Chaudhry

National Reconstruction Bureau

Social audit of governance and delivery of public services

Baseline survey 2002

National report

A Cockcroft, N Andersson, K Omer, N Ansari, A Khan, UU Chaudhry

Project undertaken with financial support from the Government of Canada

provided through the Canadian International Development Agency (CIDA)

i

Contents

CONTENTS I

List of tables ii

List of figures iii

List of acronyms iv

Acknowledgements v

SUMMARY VIII

INTRODUCTION 1

METHODS 7

Methodology principles 7

Methods in the Pakistan social audit 8 Sample and sampling 8 Data collection instruments 9 Formation and training of field teams 13 Data collection 15 Data management 17 Analysis 17 Putting the findings to work 20

FINDINGS 22

Information base 22

Overall satisfaction with public services 24 Roads and transport 24 Sewage/sanitation 26 Garbage disposal 27 Gas supply 29 Electricity supply 30

Use and experience of government health services 31 Household satisfaction with government health services 31 Access to health care 32 Payments for health services 35 Satisfaction with health service contacts 36 Monitoring government health services 38 Views of government health service providers 40

Education services 41 Household satisfaction with government education services 41 School enrolment 42 Type of school attended 43 Parental satisfaction with schools 44 Views of government school principals 45

ii

Water supply 46 Household satisfaction with government water supply 46 Type of water supply 47 Payments for water supply 47

Perceptions and Experience of the Police / Administration 49 Perceptions about the police / administration 49 Experience of contacts with the police / administration 51 Citizen monitoring of the police 54

Perception and Experience of Courts 55 Opinions about the courts 55 Experience of contacts with the courts 55 Alternative mechanisms 57

Local government 58 Voting in local council elections 2001 58 Views about new union councils 59 Contacts with union council members 60 Views of union nazims and union council members 62

Citizen Community Boards (CCBs) 64 Participation in pre-existing voluntary groups 64 Willingness to participate in CCBs 65 Community views about CCBs 66

CONCLUSIONS 69 Local issues, local solutions 71

List of tables 1. The household sample by province 2. Reported male and female household members by province 3. Costs of travel to get health care 4. Costs of travel to government health facilities 5. Costs of visits to government health facilities for a case of fever 6. Costs of visits to private qualified practitioner for a case of fever 7. Heads of government health facilities: the local government plan 8. Heads of government health facilities: effects of the local government plan 9. Reasons for children 5-9 years not attending school 10. Heads of government schools: the local government plan 11. Heads of government schools: effects of the local government plan 12. Household water access 13. Payments for water supply 14. Sources of help for a problem of personal safety 15. Sources of help for a threat to property 16. Reasons for contacts with the police 17. Reasons for satisfaction with police contact 18. Reasons for dissatisfaction with police contact 19. Reasons why the courts are there to help 20. Reasons why the courts are not there to help 21. Reasons for courts contacts 22. Reasons for satisfaction with court contact 23. Reasons for dissatisfaction with court contact 24. Reasons for expecting the new union council to be better 25. Reasons for expecting the new union council not to be better

iii

26. Reasons for contacting a union council member 27. Training reported by union nazims and councillors

List of figures 1. % households satisfied with roads 2. % households satisfied with public transport 3. % households satisfied with sewage services 4. % households satisfied with government sewage services 5. % households satisfied with government garbage disposal 6. % households satisfied with government gas supply 7. % households satisfied with electricity supply 8. % households satisfied with government health services 9. % households who usually use government health facilities 10. Proportion households who use government health services 11. % of services users satisfied with visits to govt health facilities 12. Proportion satisfied with visits to govt health facilities 13. % households who knew how to complain about govt health services 14. % households satisfied with government education 15. Net school enrolment of boys 5-9 years 16. Net school enrolment of girls 5-9 years 17. % of enrolled children attending a government school 18. % of parents satisfied with government schools 19. % households satisfied with government water supply 20. % households reporting access to government water supply 21. % households with water supply within the homestead 22. % households who would use the police for problem of personal safety 23. % households who say the police make them feel safe 24. % households with police contact in the last five years 25. % police contacts with an FIR registered 26. % police contacts satisfied with the service 27. % households who think courts are there to help them 28. % households who contacted courts in last five years 29. % households with at least one man or woman voting in 2001 elections 30. % households with at least one man voting in 2001 elections 31. % households with at least one woman voting in 2001 elections 32. Household expectations about the new union councils 33. % households in contact with union councillor, by sex of respondent 34. % households satisfied with union councillor contacts 35. % household respondents who have heard of CCBs 36. % household respondents who would be willing to join a CCB 37. % male respondents who would be willing to join a CCB 38. % female respondents who would be willing to join a CCB

iv

List of acronyms CBO Community Based Organisation CCB Citizen Community Board CIDA Canadian International Development

Agency CIET Community Information Empowerment &

Training CSO Civil Society Organisation DCO District Coordination Officer DPO District Police Officer EDO Executive District Officer FIR First Information Report GoP Government of Pakistan NGO Non Governmental Organisation NRB National Reconstruction Bureau PARAGON Regional Governance Programme for Asia PTA Parent Teachers Association SHO Station House Officer SMC School Management Committee SSP Senior Superintendent Police UNECSCO United Nations Educational, Scientific and

Cultural Organisation UNDP United Nations Development Programme

v

Acknowledgements The baseline social audit in 87 districts reported here was undertaken with the financial support of the Government of Canada provided through the Canadian International Development Agency (CIDA). The project is entitled “Citizen Community Boards: local tools for governance in Pakistan”, and also includes continuing work in one focus district. This project is part of CIDA’s Democratic Governance Programme in Pakistan. The earlier social audit in ten pilot districts is also included in this report. It was funded by the United Nations Development Programme (UNDP) in Pakistan, the UNDP PARAGON Regional Governance Programme, and the United Nations Educational, Scientific and Cultural Organisation (UNESCO), under a project entitled “Citizen participation for human rights and public service delivery at grassroots level in Pakistan”. Many people have contributed in different ways to this work. Firstly, we thank the 57,321 men and women in households who gave us information and views, the participants of some 739 male and female focus groups, the nearly 1300 head teachers and health facility heads who shared their knowledge and views with us, and the 757 union council nazims, naib nazims and members who gave us the benefit of their views and experience. In districts across the country, many people inside and outside government gave help and support to the field teams. Too numerous to name individually, we thank them all for their contribution, which greatly helped to ensure the successful completion of the field work. The field teams themselves deserve recognition for their hard and conscientious work, sometimes in very difficult conditions. Our sincere thanks go to all the women and men who made up the field teams in different parts of each province. Similarly, the data entry operators, especially those in the Karachi office, worked long and hard to enter and clean the data and we thank them for their contribution. Particular thanks are due to the members of the CIET team in Pakistan whose hard work above and beyond the call of duty has made this huge undertaking a reality within the time frame. We thank the field monitors in each province: in Sindh, Rukhsana Shaheen, in Balochistan, Farzana Baloch, in NWFP, Fazeelat Rehman and Shukria Begum,

vi

and in Punjab Dr Tayyaba Nasreen, Fozia Tanveer, Aaliya Farooqi and Nazia Gulzar. We thank Sohail Saeed, data manager in Karachi since August 2002, and his team of data entry operators for their excellent work. We owe thanks to the CIET researchers who have provided technical support for the work in Pakistan. In the 2001 pilot survey, Sharmila Mhatre led the team in Pakistan and she, Aparna Swaminathan, and Dawn Caldwell provided technical support in Pakistan, while Junko Onishi assisted with the analysis and reporting. In the 2002 survey, Alba Meneses and Manuel Pascual-Salcedo provided technical support for the first phase of data entry and trained data entry operators and the data manager. The main counterparts for this work in the federal government are the National Reconstruction Bureau (NRB). They have been enormously supportive throughout. Particular thanks go to Lt General Tanvir Naqvi, the Chairman of NRB until November 2002 and to Mr Daniyal Aziz, who was instrumental in initiating the social audit in his role as a consultant to NRB and now continues his support as the Chairman of NRB. We also extend special thanks to Mr Naeemul Haq, member of NRB, who has advised and supported us with great skill. Many others in NRB have helped with discussion and exchange of ideas. In particular, we thank Mr Anwar Butt. In each province we received help and support from many quarters and we are grateful to the many individuals and groups concerned. Some groups deserve special thanks. Sindh. We gratefully acknowledge the support of the Sindh Bureau of Statistics, especially during the ten district pilot, when they participated in design and piloting and provided logistic support for fieldwork training and data entry. We are also grateful to HOPE (Health Oriented Preventive Education) who provided logistic support for training and piloting of the 2002 survey instruments and space for data entry of records from Sindh, as well as field team members. Balochistan. We thank the provincial office of the Aurat foundation for their continuing support, including space for training and data entry. We also thank SMART and SCOPE, both NGOs, for their help and support during field work. NWFP. Peshawar University has become a valuable partner in the social audit process. Many of their senior students participated in the field work and the department of

vii

Geography and Urban Regional Planning provided logistic and administrative support for training and fieldwork. We thank the Vice Chancellor, Dr S Zulfiqar Gilani, for his continuing support. Punjab. We thank Punjab University, with whom we have developed a valued institutional link, especially the departments of Sociology, Social Work and Mass Communication, who have provided human and logistic support for fieldwork in Punjab. We offer particular individual thanks to Mr Yahya Waliullah for his help and technical inputs throughout the social audit process, first in his capacity as the Director of the Sindh Bureau of Statistics in 2001, then as the EDO IT of Karachi City District. He continues to provide support in his present role as Secretary of IT in Sindh province. The project would not have been possible without the energetic support and vision of Dr Paul Oquist of UNDP PARAGON and we thank him most sincerely. Mr M Zaheer has played a vital liaison role between the project, the donors and NRB throughout the work and we indebted to him. This report was prepared by CIET and does not necessarily reflect the opinion of NRB, CIDA, UNDP or UNESCO. CIET Islamabad, 2003

viii

Summary Social audit methodology, developed by CIET (Community Information Empowerment and Training), has been adapted to the context of the Pakistan devolution. The aims of the social audit include: • to provide policy makers with feedback on devolution

from citizens; • to provide policy makers with information about the on-

the-ground situation across the provinces; • to help district governments to make best use of their

new powers and authorities to plan and implement services tailored to the needs of their local populations; and

• to provide a means for citizens to participate in an informed way in decisions that affect their lives.

A ten district pilot of the social audit was undertaken in 2001, and data collection for a full baseline of all the remaining districts was undertaken in 2002. This report covers the findings from the full survey, including those from the 2001 pilot. In addition to the 2002 household interviews, information was also collected from service providers: 751 school principals and 310 heads of government health facilities, and from 757 union nazims, naib nazims and councillors. The field teams constructed a basic community profile for each sample community. Preliminary findings from each district were taken back and discussed in gender stratified focus groups in the survey sites, a total of 373 male and 364 female focus groups. Some 14% of households in the survey were categorized as ‘very vulnerable’ based on house construction, room occupancy and occupation of the main breadwinner. About half the household respondents were women. Overall satisfaction with public services Households were asked about their satisfaction with a range of basic public services. Satisfaction ratings varied considerably across the country. In each case, ‘very vulnerable’ households were less likely to be satisfied or even to have an available service, compared with less vulnerable households. This was partly because poor communities were less well covered by basic services and partly because the vulnerable had less access even in

Very vulnerable households A household is “very vulnerable” if it has: >Poor roof construction and >High room occupancy and >Poor occupation of main breadwinner

Household information base • 57,321 households • 433,107 people • 96,010 adult men • 87,619 adult women • 49,559 boys aged 5-12 years • 43,242 girls aged 5-12 years

Other information sources (2002) • 380 community profiles • 751 government schools • 310 government health facilities • 757 union nazims and councillors • 374 male focus group • 365 female focus groups

ix

communities where the services were available. Satisfaction ratings were related to objective evidence of services available (eg sewage and garbage). Roads and transport About a third (31%) of households were satisfied with roads, while 18% reported no roads in their area. More than half (59%) were satisfied with public transport, while 11% had no access to public transport. Sewage and sanitation Just 13% of households were satisfied with sewage services and 51% said they had no service at all. Satisfaction was related to the presence and type of sewage services present in the communities of the survey. Garbage disposal Only 6% of households were satisfied with garbage disposal and 71% said they had no service. Satisfaction and reported access were related to the presence of a system in the community. Gas supply Some 19% of households were satisfied and 71% said they had no gas supply service. Electricity supply Nearly two thirds (63%) of households were satisfied, while 16% said there was no electricity supply in their area. Health services Just 23% of households were satisfied overall with government health services, while 32% said they had effectively no access to a government health service. ‘Very vulnerable’ households were less likely to be satisfied or to report access and indeed were less likely than other households to be within 5km of a government health facility. A third (31%) of households usually used government health facilities for medical attention, while nearly half (47%) used private qualified practitioners and 21% used unqualified practitioners. In areas where there were more private services available, such as urban sites in Sindh, the less-vulnerable households were more likely to use private services, compared with ‘very vulnerable’ households who did not have a choice. All households were more likely to use government facilities if there was one within 5km.

x

Costs of travel to government health facilities were at least as high as to private facilities, on average. Costs were higher in Balochistan and NWFP, related to the distances and terrain. Costs were estimated for visits to health services in the last three months for cases of fever, in order to compare ‘like with like’. Nearly all (93%) government service users paid for visits, with a mean total cost among those who paid of Rs 232. Among users of private qualified services, 98% paid and the mean total among these was Rs 250. Some 62% of those who had used government health services in the last three months were satisfied with the service they received, compared with 80% of those who had used private qualified practitioners and 76% of those who had used unqualified practitioners. Among government-service users, those from ‘very vulnerable’ households were less likely to be satisfied with the service they received. Only 10% of households who usually used government health services knew how to complain if they were not satisfied with the service. If their usual health facility had a complaints procedure (assessed in reviews of the facilities) people were more likely to know how to complain. Education Over half of households (55%) were satisfied with the government education services in their area and only 7% considered they had no service available. ‘Very vulnerable’ households were less likely to be satisfied. Households within 1.5km of a government school (information from community profile) were more likely to consider they had a service and to be satisfied with the service. Nationally the gender gap in school enrolment among 5-9- year-olds was 13% (73% enrolment of boys and 60% enrolment of girls). The gender gap was highest in Sindh and NWFP at 18%, and lowest in Punjab at 7%. Children from ‘very vulnerable’ households were less likely to be enrolled in school, compared with those from less vulnerable households. Girls in communities within1.5km of a government girls school were more likely to be enrolled in school. The most common reason given by parents for children not going to school was that they could not afford the costs involved. Two thirds (67%) of children aged 5-9 years in school attended government schools, nearly a third (31%) attended

xi

private schools and a few (3%) attended non-formal education or religious schools. Children from ‘very vulnerable’ households were more likely to be in government schools compared with those from less vulnerable households; the difference was especially marked in urban sites of Sindh and NWFP. Some 73% of parents of children in government schools were satisfied, while 89% of parents were satisfied with private schools and 90% with the few religious schools. Parents of boys were more likely to be satisfied if the school has an active PTA but this did not affect satisfaction of the parents of girls. Parents of girls were more likely to be satisfied if the school provided free textbooks for girls. Water supply Only 18% of households nationally reported satisfaction with government water supply and 62% said they had no access to government water supply. ‘Very vulnerable’ households were less likely to be satisfied or to have access to government water supply. Overall, 78% of households had a water supply within the homestead, but this figure was only 58% in Balochistan. ‘Very vulnerable’ households were less likely to have their water supply within the homestead. About a quarter (26%) of households used a piped water supply (inside or outside the homestead), 66% used some form of ground water, 5% used surface water and 2% got water from tankers or vendors. Four out of ten (44%) households paid for their water supply, with a mean among those who paid of Rs 362 per month. Police When asked who they would contact for a problem of personal safety or a threat to property, the most common response of households (more than a third) was “Allah”. Some 22% said they would contact the police for a problem of personal safety and 25% for a threat to property. The proportions who would contact the police (or levies) were particularly low in Balochistan. Male respondents were more likely to say they would contact the police. Households in communities with a police station were more likely to say they would contact the police. Nearly a third (30%) of households said the police made them feel safe. Women and respondents from ‘very

xii

vulnerable’ households were less likely to say the police made them feel safe. Households with a police station in the community were more likely to feel safe through the police. Only 12% of households reported contacts with the police over the last five years. Households in communities with a police station in the community or within 5km were not more likely to report police contacts. Male household respondents were much more likely to report household police contacts compared with female respondents, who were apparently not aware of all contacts. ‘Very vulnerable’ households were less likely to report police contacts. About half the police contacts were initiated by the police rather than the households. The proportion of contacts initiated by the police was higher for ‘very vulnerable’ households. In about half the contacts (49%) a First Investigation Report (FIR) was registered; this was more likely when the contact was initiated by the household rather than by the police. A third (32%) of households who contacted the police were satisfied with the service. Households who initiated the contact themselves were more likely to be satisfied. Courts Some 46% of households said they thought the courts were there to help them. Many people, especially women, were unable to give a definite response to this question. Only 8% of households reported a contact with the courts in the last five years. Male household respondents were more than twice as likely to report a court contact, suggesting women were not necessarily aware of such matters. ‘Very vulnerable’ households were less likely to have had contact with the courts. Households within 10km of a court were somewhat more likely to have had a court contact. Half (49%) the households who had contact with the courts were satisfied with the experience. ‘Very vulnerable’ households were less likely to be satisfied. Few (7%) households had heard of the alternative reconciliation committees at union council level. Awareness of these committees was higher among urban households but lower among ‘very vulnerable’ households.

xiii

Local government Nationally, 89% of households reported at least one male member voting in the 2001 local elections, and 72% reported at least one female member voting. The gender gap in voting was especially marked in NWFP, which had the highest proportion of male voting but the lowest proportion of female voting. ‘Very vulnerable’ households were less likely to have at least one woman voting. A third (33%) of households thought the new union councils would be better than the previous system, 35% thought they would not be better and 31% could not say if they would be better or not. Some 21% of male household respondents reported a household contact with a union councillor, but only 10% of female household respondents reported such a household contact. This suggests women were not always aware of such contacts made by household members. ‘Very vulnerable’ households were only marginally less likely to report a contact with a union councillor, but the difference was more marked in rural areas. Just over half (54%) the households who reported a contact with a union councillor were satisfied with the contact. ‘Very vulnerable’ households were less likely to be satisfied, especially in urban areas. Citizen Community Boards Few households reported membership of existing voluntary groups: 2% had a male member of a group and 0.6% had a female member. Almost no households without a man in a group had a woman in a group. ‘Very vulnerable’ households were less likely to have a member in a group. Very few household respondents had heard of CCBs at the time of the data collection: 3% of men and 1% of women. Once they had heard a brief explanation about CCBs, 50% of men were willing to join a CCB, but only 29% of women. The gender gap in willingness to join a CCB was most marked in NWFP, which had the highest proportion of men interested in joining a CCB (66%). Respondents from ‘very vulnerable’ households were as likely to be willing to join a CCB as those from less vulnerable households.

1

Introduction Devolution – the local government plan In an effort to improve both governance and democracy, the Government of Pakistan through its National Reconstruction Bureau (NRB) took a major initiative by introducing the new local government system. The local government plan was announced in October 2000. The initiative set up structures to devolve authority and decision making capacity to lower levels of government and community. Elections to the new union, tehsil and district councils took place in 2001. The new system replaced one that had been in place, more or less in the same form, since the 1800s. The devolution reform is intended to improve access to public sector services, to encourage sustainability of local development initiatives and to add to public sector resources through community mobilization of resources and through increased transparency and reduced leakages of resources out of the system. The reform combines devolution of political power, decentralization of administrative authority, deconcentration of functions, redistribution of resources, and checks and balances intended to diffuse the power-authority nexus. In the new dispensation, empowerment of citizens is envisaged through officially recognized Citizen Community Boards (CCBs) and enhanced representation through elections to the three levels of elected local government: unions, tehsils and districts. Monitoring devolution A system for collecting detailed institutional data and using this to monitor the working of the system has been set up: the National Reconstruction Information Management System (NARIMS). This new system is being fine-tuned in several districts at present. This will provide district, provincial and federal government with much-needed information about how the systems are working. It will also hopefully address an important problem with presently available institutional data: the unevenness in the reliability of the information that makes it all but useless for the purposes of evaluating performance or allocating resources from a central level.

2

However, collecting good quality institutional data is not the whole story, even in an improved form. It does not ensure that the citizens’ voice and priorities are involved in the decision making that affects the public services they can access and the quality of their lives. Institution based information includes only those people who have access to services. People who do not attend government health clinics, schools or police stations are not part of the system. Yet those people excluded by the system are often the most disadvantaged members of society, most in need of the services. This problem can be addressed by a district level survey process designed to include the most vulnerable groups. The skills to undertake such a process are currently not available at district level. Most large scale survey processes are outside the local skill set. But stand alone externally managed monitoring mechanisms run the risk of encountering resistance or neglecting local sensitivities, resulting in lack of ownership of the results. Local planners not only need to know about the current situation in their own locality, including the views of citizens. They also need to know about what solutions or interventions are likely to work best in their area, in order to invest resources accordingly. This is evidence-based planning and it implies an analytical capacity that requires substantial training. Data do not, even when relevant, timely and accurate, “speak for themselves”. As well as information for planning at local level, there is a need for a feedback mechanism to policy makers that provides information about citizen priorities, views and experience of public services and involvement in local governance. Such information allows the tracking of the effects of reform over time and across the country, so that policies can be differentiated by territory and modified over time. Citizen involvement under devolution An important component of the devolution initiative is that is has established structures and processes intended to promote community participation and monitoring. Empowerment of citizens hinges through the new Citizen Community Boards (CCBs). The CCBs are intended to gather together community views on human rights concerns, citizens’ security and social service delivery, monitor government operations, make recommendations regarding government policies and practices, and spearhead self help practices and projects.

3

Even more than local government officials, members of the new CCBs will need to develop their skills to monitor public services in their locality, and to develop proposals and implement development schemes. Their involvement in systems set up to monitor devolution, public services and local governance should be designed so that it enhances their skills and confidence to make the most of their newly designated rights and powers. The social audit and devolution The social audit methodology, developed by CIET1 (Community Information Empowerment and Training), over two decades and across more than 40 countries, has been adapted to the context of the Pakistan devolution. Implementation of the social audit has several purposes in relation to devolution. First, at the national policy level, the social audit provides a citizen feedback channel. It provides policy makers with a means of tracking the views of citizens about public services, and their use and satisfaction with available services. In this sense, it provides the “bottom line” about service reforms from the viewpoint of the intended beneficiaries. It is also a means of measuring citizen knowledge of and participation in local governance mechanisms, including voting in local elections, use and views of structures such as reconciliation tribunals, and membership of CCBs and other community organizations. As the devolution reforms progress over time, there should be an improvement in delivery of public services and an increase in citizen satisfaction, as well as increased engagement of the public in bodies such as CCBs. The social audit, repeated regularly, is a key means of checking whether the reforms are having the desired effect, providing guidance about areas where there are problems and suggesting ways to fine tune the process to increase its effectiveness. Second, at provincial level, the social audit provides policy makers with information about the on-the-ground situation across the province. Such information is often lacking and this seriously hampers policy makers trying to decide about the most efficient and equitable allocation of financial resources to districts, as well as the most effective way for them to provide technical support for district governments. Over time, repeated social audits provide the provincial policy makers with a way of checking the implementation

1 CIET is an international group of non-profit NGOs, academic institutes and charities dedicated to building the community voice into planning.

“The social audit will provide a comparison between what should have happened and what actually happened” Gen Naqvi, Chairman NRB, November 2002

4

of policies across districts and the effectiveness of implemented policies in terms of citizen perceptions, use and experience of public services. Third, the social audit, especially once established at district level, is intended to help district governments to make best use of their new powers and authorities to plan and implement services tailored to the needs of their local populations. The social audit at district level involves district officials in designing and implementing data collection from all segments of the population in their district, and in discussing the findings with community representatives to plan service improvements. This helps to build the skills they need to plan and deliver services that best suit the needs and aspirations of their district. Last but certainly not least, the social audit at district level and below provides a means of giving form to the intention under the devolution initiative of empowering citizens to participate in an informed way in decisions that affect their lives, and encouraging them to engage in local democratic processes. Over time, the social audit process will involve members of CCBs (and other community bodies) in collecting information about priority issues, and using this to help them formulate development proposals for funding from local government under the cost-sharing arrangements that are part of the local government plan. In the initial phase, the social audit requires substantial outside technical support. Collecting reliable information that reflects the whole population, including vulnerable groups, and analyzing it in a way that supports decision making, is not an easy process. However, over time, it is important that skills for the process can be extended down to local levels, as well as becoming embedded in the routine procedures of government planning at all levels. Steps to ensure this sustainability are an important and integral part of the social audit. This includes training government and other people in the social audit methods, both through their involvement in cycles of data collection, analysis and use, and through provision of formal training courses in the methods. It also means that the social audit should be tailored to fit in with existing systems and procedures as far as possible. For example, training of district government officials and elected representatives in issues of evidence-based planning related to the social audit is proposed to be undertaken as part of the training being provided for these same people by NRB and others. And the information from surveys at district level should be used in planning not as a

“At present, we have to plan based on our own perceptions. Having proper information and evidence will help us to plan sincerely and allocate resources properly to improve the quality of life for all our citizens.” Zila Nazim, Lasbela April 2003

5

separate exercise, but as a part of the district planning and budgeting cycle. Implementation of the social audit in Pakistan In 2001, the United Nations Development Programme (UNDP) in Pakistan, the UNDP PARAGON Regional Governance Programme and the United Nations Educational, Scientific and Cultural Organisation (UNESCO) provided funding for an initial ten district pilot social audit. This project was entitled “Community monitoring of public services and human rights in Pakistan”. The intention was to establish the feasibility of the process and to establish a basis for extending the social audit to the whole country and in more depth. In 2002, the Canadian International Development Agency (CIDA) provided funding for a project entitled “Citizen Community Boards: local tools for governance and community monitoring in Pakistan”. Under this project, CIDA provided funding to extend the baseline social audit data collection to all the remaining districts of Pakistan, as well as to undertake detailed work to establish the social audit process at district and sub-district levels in one focus district. The lessons learned about establishing the process at district level in the focus district can then be applied in extending the district social audit to all districts. The overall aims of the CIDA funded Democratic Governance Programme, of which the CIET social audit is one element, are: 1. Improved local governance policies and policy

implementation 2. More effective local democratic institutions and

practices 3. More effective participation of women in local

governance, and 4. More effective citizens’ voice in setting local priorities

and delivering social services and access to justice. The 87 district survey undertaken in 2002 and the survey in 10 districts in 2001, together provide a full national baseline of the situation at the beginning of devolved local government. This can be used as the basis for tracking over time changes to public perceptions about delivery of public services delivery and local governance. Annual surveys are planned for the next five years Technology transfer during this time will institutionalize the social audit methodology in Pakistan.

Social audit baseline survey 2001: ten districts pilot 2002: all remaining districts

6

Reporting on the social audit Reporting on the social audit will take place at a number of levels and by different means. This document is the formal report on the baseline social audit, incorporating the findings from the 2001 ten district pilot survey with the 2002 survey of all the remaining districts. It is planned to use summaries from this report for dissemination and publication of the findings through appropriate channels. The findings in this report will also be of interest to the four provincial governments, enabling them to see their own findings in relation to the other provinces and in the overall national context, as well as giving them information about the spread of findings across their own province. Individual district reports based on the findings have been prepared, summarizing the findings in the district and showing them in relation to the relevant province, as well as in the national context. These reports will be used as the basis for discussions about the social audit and the findings of the baseline survey with district governments. In the social audit reporting of findings is not the end of the work, but rather the beginning of the next phase: discussing the findings with concerned constituencies, from government, from civil society, and from other stakeholders, such as donor agencies. This report is a part of the discussion process at national level. The process will also include the media, for example through the media and governance initiative of the NRB. In the focus district for the CIDA-funded project, Lasbela in Balochistan, the district findings have already been discussed with the district elected representatives and officials. Reports of the district findings as part of the national baseline survey were used as the basis for discussions about the findings and the ongoing social audit process at union council level.

7

Methods

Methodology principles The CIET social audit methodology has been developed over two decades, working in over 40 countries world wide. The methods are described in detail elsewhere and an overview of the methods is included as Annex 1. Social audits increase the informed interaction between communities and public services. A unique combination of quantitative (survey) and qualitative (key informant and focus groups) evidence focuses on the impact, coverage and costs of public services. Civil society, drawn into interpretation of these data in an ordered manner, plays the pivotal role in building local solutions. This democratises the decision-making processes and includes the voice of the people in planning. At the national level, these representative data and human interactions around the data provide a step-ladder to increased accountability in public services. The originality and contribution of CIET methods lie in (i) the incorporation of modern epidemiology to evaluate evidence for planning and (ii) the fact that the community voice plays a central role in that evidence, its analysis and resulting action. Accuracy of decisions that result from the use of epidemiological method gives meaning and volume to the community voice, increasing the confidence of civil society in its participation in governance and thus service reform. The concept of the social audit is simple: collect information about public services from people supposed to be served, and from service providers, and use this as a basis for involving the public and service providers in making changes to improve the services. The key steps include: collect information from households in representative communities about their use, experience and perceptions of public services; link this with information from the services themselves; analyse the findings in a way that points to what actions might improve matters; take the findings back to the communities for their views about what could improve the situation; bring the findings and suggestions to discussions between service providers, planners and community representatives to plan and implement changes. The loop is closed when a repeat fact-finding exercise assesses the changes and their effects.

A social audit cycle -Clarify the strategic focus -Design instruments, pilot test -Collect information from households on use and perceptions of public services -Link this with information from the public services -Analyse the findings in a way that points to action -Take findings back to the communities for their views about how to improve the situation -Bring evidence and community voice into discussions between service providers, planners and community representatives to plan and implement changes

8

Methods in the Pakistan social audit

Sample and sampling In the 2001 ten district pilot social audit, two districts each were purposively selected from Sindh, Balochistan and North West Frontier Province (NWFP) and four districts from Punjab. In 2002, the remaining districts in all four provinces were included in the sample. In Balochistan, some splitting off of districts took place after the sampling and data collection, so in the analysis for Balochistan, these ‘new’ districts are included with the district existing at the time of the sampling and data collection. Within each district, representative communities were selected by a two-stage stratified random sampling process. The sampling frame in each district was the official list of union councils within the district. The union councils on the list were first stratified into rural and urban types, according to official definitions. The proportion of urban and rural sites to be included in the sample was set according to the urban and rural population proportions in the 1998 census. The allocated number of union councils for the district was then picked randomly from the lists of urban and rural union councils for the district. For the ten districts in the 2001 pilot, six union councils were randomly selected per district. For the remaining districts included in the 2002 survey, a minimum of four union councils was randomly selected per district. For particularly populous districts, more union councils were selected. So, for example, up to eight union councils were selected for districts in Punjab. In Karachi, the interest of the new union councillors in the process allowed more coverage than originally budgeted, and it was possible to select randomly at least one union council in every town of Karachi City District, giving a total of 20 union councils, including the six covered in the 2001 pilot. For each of the randomly selected union councils we obtained the list of communities or villages and made a random selection of one community from each list. In each selected community or site, all contiguous households up to around 120 were included, radiating from a randomly allocated fixed point in the community. There was no sampling within this site: all households were included in the sample. We deliberately included at least four union councils per district, even for sparsely populated districts, in order to

Districts in the 2001 pilot Sindh Karachi West Thatta Balochistan Jhal Magsi Kalat NWFP Kohat Mardan Punjab Jhelum Mianwali Narowal Rawalpindi Districts in the 2002 survey 89 districts *Rest of Karachi included *Two districts in Balochistan were subdivided after the data collection

The social audit sample sites

9

collect enough information from every district to be able to analyse and present the information of every district separately, rather then just “as a part of” a wider provincial picture. This is important in the process of giving back the information and discussing and using it at district level. Because of the wide disparity in district populations across the country, this inclusion of at least four sites per district resulted in over-sampling of most districts in Balochistan and NWFP and under-sampling of most districts in Punjab (even though more than four union councils were selected from in the bigger districts), relative to their proportions in the actual population of the country. In order to take into account this disproportion in the sample population distribution, we calculated a weight for each district according to its relative over or under representation in the sample. All the indicator percentages mentioned in this report at provincial or national level are the weighted values, unless stated otherwise.

Data collection instruments As mentioned in the general description of the CIET social audit methods, an important feature is the combination of quantitative and qualitative data, collected from the same places around the same issues. This allows the qualitative data to be used to give context to the quantitative findings and to be used in combination with the household and other quantitative data in a process of mesoanalysis. Therefore, a number of different instruments were designed and used in the baseline survey of the social audit. In 2001, CIET, in consultation with key stakeholders and partners, including the NRB, UNDP and UNESCO, developed the instruments for the ten district pilot survey. CIET instrument design takes a standards-based approach, using previously validated questions from previous CIET surveys and others whenever possible. In this case, since this was to be a baseline survey, it was decided to include a number of key public service sectors, as well as issues about local government, even though this would limit the amount of detail that could reasonably be collected about each sector. The instruments were piloted as part of the design process: first they were tested within the design group, then tested on immediate contacts, then individual questions or sections were piloted in a community setting, then the full instruments were tested in non-sample communities, and finally the data entry and analysis formats developed as part

10

of the design were tested using data from the pilot data collection exercises. This piloting is a standard part of the CIET instrument design and aims to ensure that all the questions have a clear purpose – they produce information that will be used in the analysis – and that all questions needed for a meaningful analysis are indeed included (for example, questions about factors that could explain differences in access to services are included). The data collection instruments were translated into Urdu, with back-translation to check meanings had been retained, and subsequently translated into Sindhi, Balochi, Brahvi, Pashtoo and Punjabi for their administration in different areas of the country. In some communities the interviewers carried the instruments with them in more than one language. In 2002, CIET and the design group (including representation from the NRB) reviewed the instruments in the light of the 2001 ten district pilot exercise and made a number of changes to improve and extend the information that could be collected. The same basic questions were retained so that the information from the ten pilot districts could be combined with that from the remaining districts to give a complete national baseline from all districts. The household questionnaire This is the main quantitative data collection instrument. In this case, the questionnaire comprised several sections: A general section covered house construction and demographics of the household members, including education and occupation of the main breadwinner. Sections on public services enquired about perceptions, use and experience of water supply, health services, education, police and courts. In 2002 a section about overall satisfaction with a range of public services was added and the section about primary education was revised for clarity. A section about local government asked about intention to vote or voting in the union council elections (some communities were surveyed just before the 2001 elections, some just after) and expectations about the new councils. In 2002, the questions about voting were more specific and questions about contacts with the new union councillors were added.

Registers for household data collection

11

A section about community participation asked about membership of voluntary groups and knowledge about Citizen Community Boards, then gave brief information about CCBs and asked about willingness to participate as a CCB member. In the social audit, both the 2001 ten district pilot and the 2002 main survey, the household questionnaire was administered using a standard CIET method whereby the two halves of the questionnaire are attached to the inside cover of a register and the responses to each household are recorded on a separate, numbered page of the register. Each interviewer uses one register to record the responses for all the households she or he interviews in a given site. This method is cost-effective and prevents loss of or damage to individual completed questionnaires in field conditions. It also facilitates data entry, retrieval of particular records during data checking, and storage of the original data records. The community profile questionnaire This instrument collected information about features of the community that could be relevant to the use and experience of public services, such as the types and locations of health and education facilities, garbage arrangements and the availability of radio and newspaper services and community based organisations. This instrument was completed in discussion with a community leader, contacted in any case as the field team entered the community to conduct the household survey. The community profile instrument was refined for the 2002 survey, in order to ensure it adequately covered issues that could be linked to responses to the household questionnaire. Key informant interview schedules with service providers These instruments were used to collect information from school principals and heads of health facilities about their facilities. In addition, some sections were completed based on observations the interviewers made in the facilities. The information included issues likely to be relevant to the level of use and to the experience of service users. For schools, the information included class size, staff/pupil ratios, and facilities such as electricity and water supply, classroom furniture and equipment, toilets, and boundary walls. For health facilities the information included staffing, official charges, complaint systems, health education arrangements, and observation of facilities.

Inside of a household data collection register

12

Again, these instruments were revised in the light of the 2001 pilot: some questions that proved to be not very useful were removed and additional relevant questions were included and piloted. Key informant interview schedules with union councillors In those communities in the 2001 ten district pilot that were covered before the union council elections a community elder or a candidate standing for election to the union council was interviewed. In those covered after the elections in 2001 and in all the remaining districts covered in 2002, members of the new union councils were interviewed. In 2002, we tried to interview the union nazim or union naib nazim in all the union councils in the survey (at least four per district), as well as another council member and specifically a female council member. We made particular efforts to contact a female union council member in each case. The interview schedule, revised for the 2002 survey, included questions about priority problems in the union council, the information needs of council members, their methods of seeking the views of citizens, issues about financial support, and knowledge and views of CCBs and their formation in the union council. Focus group guides In the CIET social audit methodology, focus group discussions serve several purposes: to collect qualitative information to complement the quantitative information collected through the household survey and key informant interviews with service providers; to begin the process of feeding back the information from the survey to the participating communities; and to seek their views about how important information from the survey findings could be effectively communicated to people such as themselves. In the 2001 ten district pilot survey, two sets of gender stratified focus group discussions in the sample communities were conducted: one at the time of the household data collection and one later, when some key findings from the household survey were discussed. The initial focus groups included issues about community information needs and views about potential CCBs. In the 2002 survey, we conducted only focus groups to discuss the household survey findings.

13

The focus group guides are not questionnaires; they are topic guides to ensure that the discussion is indeed focused and to share findings. The guides for the feedback focus groups were therefore developed once the basic analysis of the household survey had been undertaken. For the focus groups in each district the relevant findings from that district were included in the guide. The feedback focus group discussion guides in the main 2002 survey covered the topics of: priority public service problems; the potential for CCBs in the community, the difficulties for setting up and suggestions for how CCBs could work effectively; the satisfaction with health services and how CCBs could monitor health services; whether the police made people feel safe and how citizens could monitor the police.

Formation and training of field teams In the both the 2001 ten district pilot survey and the 2002 main survey, field teams were recruited and trained in each province: in 2002 each province was divided into regions for the purpose of the field work. People recruited into the field teams included government employees (for example, teachers, bureau of statistics staff), members of NGOs and other Community Based Organisations (CBOs) and senior university students as well as recent graduates. Some of the people recruited had worked with CIET on earlier projects, while others were new to the process. Teams for the household survey Training for the household survey and key informant interviews took place over three or four days, depending on the progress of the group. In all training sessions, rather more people joined the training than were required for the field teams in the area, and only those who did well in the training were selected for the teams. The training included classroom sessions to give the teams an understanding of the purpose of the social audit and a basic understanding of the CIET methods. Also in the classroom the trainers went through the household questionnaire in detail. The team members responsible for interviewing key informants in the schools and health facilities and union councillors received additional training about the data collection instruments for these parts of the survey. The trainees practised administering the questionnaire to each other in the classroom, including role-playing the introductions and potential difficult scenarios. One half day of training was spent on a field practice in a nearby non-sample community, with close monitoring and feedback about any

Role play during training Photo: CIET

Classroom training Photo: CIET

14

errors and misunderstandings. The final day of the training was a full mock data collection in the field, with the proposed teams working together. If the trainers were not satisfied with the work in the mock data collection day, the training was extended with further feedback and further field practice until the trainers were satisfied. Each field team comprised three female interviewers, two male interviewers, one male logistic control associate and one female quality control associate. Training took place in each district of the ten districts in 2001 and in regional centres in 2001. In 2002, three training sessions took place in Sindh (one shared with Balochistan), one more in Balochistan, four in NWFP and four in Punjab. Some 200 people were trained for the 2001 pilot and over 500 were trained for the 2002 fieldwork. Teams for the focus group discussions Field teams to conduct the focus group discussions, returning to the same sites as for the household data collection, comprised two men and two women for each site. Usually two teams travelled to a district together, making a total of eight: four men and four women. In nearly all cases, at least one of the people returning to a community to conduct focus groups had been a member of the team doing the household survey in the site, so was already familiar with the layout and had contacts in the community. More than half of the team members trained to conduct the focus group discussions had participated in the household data collection. The training for the focus group teams took place over two to three days, depending on progress. On the first day the trainers explained general principles about facilitating and recording focus group discussions and went through the guide for the focus groups to be conducted. The trainees conducted focus groups among themselves in the classroom, monitored by the trainers. The next day the trainees conducted focus group discussions in a nearby field site, again closely monitored. A further day’s practice was included if necessary.

Focus group training Photo: CIET

Focus group training Photo: CIET

15

Data collection The field teams undertook household data collection for the ten district pilot survey in May to July 2001. After some basic preliminary analysis, teams returned immediately to the same communities and conducted focus group discussions. The training of field teams and household data collection for the main survey of all remaining districts in 2002 took place according to a phased schedule, moving from province to province. The first province to be covered was Sindh, followed by Balochistan, then NWFP and then Punjab, with some overlap between provinces. In each province a CIET provincial coordinator has responsibility for the activities in that province. The Sindh provincial coordinator is also the national coordinator. The provincial coordinators scheduled the field work in each province, such that coordinators were able to help each other during training sessions. In general, during the main 2002 survey each field team could complete the household survey in a community in one day and move to another community the following day. In some cases this was not possible due to terrain or difficult weather conditions and additional time was allowed for travel or to complete data collection from a site. For both the 2001 ten district pilot survey and the main 2002 survey, the field teams took with them a letter from the NRB, giving official status to the work. The teams also made contact with district government officials as they entered each district, to brief them about the aims of the social audit and to seek their support for the field work as necessary. In some districts in particular where the security situation was difficult, the district officials were very helpful to the teams, giving them advice and sometimes logistic support to reach difficult communities. In 2002, the field teams managed to collect data from all the remaining districts, even those with marked security problems. They achieved this using good local knowledge and contacts, as well as with support from district government and administration. On entering each community, before the team started the household interviews, the logistic control associate made contact with community leaders to explain the purpose of the survey and seek their support to work in the community. In nearly every case the community leaders agreed to the survey. In

An interview in an urban site Photo: CIET

An interview in a rural site Photo: CIET

16

the handful of communities (less than ten in total) where the leaders refused to allow the survey to take place, the field team undertook the survey in the next neighbouring community instead. The focus group discussions that were part of the 2002 main baseline survey process took place between the autumn of 2002 and the spring of 2003. The focus group guides included, for each district, some key findings from that district based on the basic analysis of the data from the household survey. Again, the field teams managed to conduct focus group discussions in all districts, even some with serious security problems, because of good local involvement in the field teams and good local knowledge and contacts. In most communities, the teams conducting the focus groups were warmly welcomed back, as they had “kept their promise” to return to the community to share and discuss some findings. Quality control of data collection Quality control during fieldwork is a crucial concern. In the social audit baseline surveys it was ensured in several ways: >Careful training emphasized the importance of proper conduct of the household and other interviews, according to strict guidelines, leaving no room for individual interpretation by interviewers. >Only those trainees who showed themselves capable of good, careful work were selected for the field teams. >All team members were told that if they did not work properly while in the field, they would be asked to leave the team immediately. On rare occasions in the course of the data collection of the social audit it was necessary to dismiss field team members and replace them with reserves. >In each team, the quality control associate checked the registers completed by the interviewers, first after they had completed three households, then in the middle of the day and at the end of each day. She pointed out any incorrect recording of information and instructed the interviewer to return to a household to collect missing information if necessary. The quality control associate also randomly visited some households to check they had actually been interviewed. >The field coordinators and provincial coordinators between them visited the field teams in the field sites to

Field check of registers Photo: CIET

A provincial coordinator checks a register Photo: CIET

“Today the well has come to the thirsty ones.” Male focus group, Panjgur (on welcoming the focus group facilitator and recorder)

17

check how they were working, to deal with any identified problems and to make a further check on the work of the quality control associates within the teams.

Data management Data entry for the 2001 ten district pilot survey took place in Karachi in the offices of the Sindh Bureau of Statistics. For the main 2002 survey, the data for Sindh were again entered in the premises of the Sindh Bureau of Statistics. For Balochistan training for data entry and data entry took place in Quetta, in the offices of the Aurat Foundation. CIET then set up a specific data management unit in Karachi, and the remainder of the data entry, from NWFP and Punjab, all took place in this facility. CIET trained all data entry operators and selected the best candidates to undertake the work. In Sindh, many of the operators have worked with CIET on previous projects and are very experienced in the data entry methods used in CIET work. Data entry was programmed using the public domain epidemiological and statistical software package Epi Info, version 6. All data were entered twice and validated using the Epi Info Validate facility. Double data entry and validation greatly reduces key stroke errors in the dataset. After validation, further cleaning of the dataset looked for logical errors, out of range responses and duplications. The cleaning was completed by checking back to the original data registers as necessary.

Analysis Analytical approach Annex 2 gives a description about CIET levels of analysis. The full analysis goes beyond the calculation of frequencies of indicators. While levels of indicators describe the present situation – and it can be of interest to look at their variation across the country, for example – this is not of much help to planners who need to develop strategies to change the situation. Further analysis looks at the actionable factors that are related to the important indicators. The potential affects of changing these factors on the outcomes can be calculated. In doing this, it is important to take into account the other factors that might be the real cause of apparent associations (confounders) as

Data entry operators at work Photo: CIET

18

far as possible, otherwise spurious associations can be misleading. The CIET analysis complements quantitative data analysis with semi-quantitative and qualitative elements from key informants and focus groups, using meso-analysis. For example, sites can be coded to reflect the distance of the government health facility from the community, or the views of the union nazim about the formation of CCBs. Focus group discussions generate richer and more textured evidence than structured interviews with key informants or individual quantitative questionnaires. Information in focus group reports can be coded to reflect certain views (for example, interest in forming CCBs) that may be present in some communities but not others. This coded information can then be linked to the records of individuals from the communities, using the community code as the link. Meso-analysis essentially deals with factors operating in the community or peer group by linking them to the experience of the individuals in that community. Mapping The CIET mapping approach is also described in more detail in Annex 2. The maps produced are raster maps, with interpolation of levels of indicators between the sample sites. Importantly, the weights for the sites in relation to their population contribution are taken into account in constructing the maps, so that the area of the map in a particular colour represents the proportion of the population with that level of the indicator, and not just the geographical distribution. The maps should be interpreted essentially as weather maps, with the focus on the overall picture rather than on individual positions on the map. For the social audit baseline survey, we constructed maps that combined the findings from the ten district pilot survey with those from the full survey of the remainder of the districts. For variables where there was no result from the ten district pilot (because the question was added after the pilot) the ten districts a shown as grey, meaning “no data”. Analysis of the baseline surveys We combined the data from the ten district pilot survey with the data from the survey of the remaining districts to give a dataset covering all districts of the country. Some districts in Balochistan have been sub-divided since the data collection; in these cases the data from the ‘parent’ district cover all the subdivided districts.

Proportion of households who were satisfied with the roads

CIET2003

19

Weighting As mentioned in the section on sampling, the deliberate inclusion of an adequate sample from every district resulted in over-sampling of sparsely populated districts of Balochistan and NWFP and under-sampling of densely populated districts of Punjab. This was taken into account by calculating weights for each district (the ratio of their fraction in the sample population to their fraction in the actual population). Annex 4 shows the district sample populations and the district census populations and the weights calculated for each district and province. All the main indicator percentages quoted in the report for provinces or at federal level are quoted as weighted values. Some questions were added in the main survey based on the experience of the ten district pilot survey. In the responses to these questions, there is therefore no data for the ten districts covered in the pilot survey. Epidemiological analysis The statistical and epidemiological analysis of the findings from the baseline survey was undertaken using the Epi Info statistical software package. In the analysis of each sector, promising associations between actionable factors and outcomes of interest were analysed using standard epidemiological techniques to identify potential effect modification or confounding by factors like socio-economic status, gender etc. The effects of these factors were examined using stratification and the Mantel-Haenszel procedure. Contrasts and associations are reported mainly in terms of the Odds Ratio (OR). For example, a man was about twice as likely to be willing to become a member of a CCB, compared with a woman (the OR here is 2.0). The confidence intervals around the Odds Ratio are given to indicate both the accuracy of the estimation and the likelihood that it could be explained by chance alone. If the 95% Confidence Interval of the Odds Ratio does not cross 1.0, then the likelihood that the association is due to chance is less than 5%. The confidence intervals quoted are those of Cornfield. Heterogeneity between strata in stratified analyses, indicating effects in certain subgroups and not others, was formally tested for using the Woolf procedure. It is always possible that some of the apparent associations reported from this survey are actually due to unknown and unmeasured confounding factors. However, by excluding most of the important likely confounders, then the findings do offer a starting point for action.

20

Vulnerability analysis A key intention of the devolution process is to improve the lot of the most disadvantaged members of society, enabling them to have a say in decisions affecting their lives and ensuring they are better served by public services. As part of the analysis of the baseline survey we identified vulnerable groups and looked at their situation compared with that of the less vulnerable. We looked at two vulnerable groups in particular: women, and the most vulnerable households in a socio-economic sense. We defined household vulnerability in terms of three factors: house construction, room occupancy and occupation of the main breadwinner. A category of ‘very vulnerable’ according to all three factors was defined that included some 14% of the sample population nationally, with considerable variation between areas of the country. We used this to examine the situation of the most disadvantaged section of the population compared with the rest, for access to, experience of and satisfaction with public services.

Putting the findings to work This report is not intended as the conclusion of the work of the first cycle of the social audit. Rather it is the raw material for generating discussions about the findings with key players at all levels. The discussion material will be prepared based on the facts and analysis in this report, tailored to the particular needs of the audience being addressed. Following the analysis and reporting of the 2001 ten district pilot survey, CIET facilitated discussions about the findings in all the participating districts. In all cases a workshop was held at district level, organised by CIET and the district government. Participants included the district elected representatives, the district officials, representatives from tehsil and union governments, and representatives from civil society organisations (CSOs) and NGOs. In some districts, the findings were first discussed by stakeholder groups to consider potential action plans arising from the evidence. They then presented their conclusions at the district level workshop. In Karachi, for example, where Karachi West was included in the 2001 ten district pilot survey, the workshop in Karachi City District heard presentations from the police,

21

education and health departments about the findings from Karachi West and their plans about improving the situation. Partly as a result of the interest generated around the findings during this workshopping process, a number of the new councillors (especially women councillors) came forward to assist with the survey in the rest of Karachi in the 2002 national survey. This allowed us to increase the number of sites from Karachi included in the survey to include one from each town. Reports of the discussions in the ten districts are available. But this is only a small element of what is needed to make the most use of the evidence from the social audit. CIET has described a process of socialising the evidence from social audits for participatory action, which includes a number of elements. In Pakistan, this socialisation of the evidence from the baseline social audit national survey is just beginning. In one focus district, the full process is being demonstrated, learning lessons and producing a “toolkit” that can help to extend this process to all other districts. The focus district is Lasbela in Balochistan. The government of Balochistan has given support at the highest level support for the work in Lasbela over the next three years. The CIET team has discussed the findings from Lasbela in the national baseline survey with the district government and elected representatives from tehsil and union governments. This process continued with discussions about the findings held at tehsil and union levels, including members of the existing CCBs in the district. District government focal points and other stakeholders participated in design for more detailed data collection in the district, to reflect the diversity of conditions across the district and to include issues of particular relevance to Lasbela. Data collection has been completed and preliminary analysis will be completed in the autumn of 2003. The evidence from this exercise will be widely shared and discussed in the district and used as a basis for supporting the emerging role of CCBs in the district as well as supporting and beginning to build capacities of district officials for evidence-based planning.

District government, Lasbela, consider the findings of the social audit Photo: CIET

22

Findings

Information base Household population The total national survey (including the 2001 pilot) in all districts of Pakistan covered some 57,321 households (433,107 people) in 444 representative communities. In the 2002 survey, the household survey covered 384 communities. As well as the household survey, information was also collected from other sources: community profiles, schools, health facilities, union nazims and councillors, and male and female focus groups. The number of households in the sample in each province, the household size and the total population covered in the households are shown in Table 1. There was clear evidence of under-reporting of female household members, both adult women and girl children. This was apparent across all the provinces (Table 2) and is also a feature of the 1998 population census and of other surveys in Pakistan. The respondent answering on behalf of the household was a woman in more than half the households (55%, n=57294). Less than half of them (46%, n=57182) had any education. Male respondents were nearly five times more likely to have some education compared with females2. Most (82%, n=57311) household heads were male. Around half of them (51%, n=57223) had at least basic education. The proportion of household heads with no education was highest in Balochistan at 68% (n=12225) and lowest in Punjab at 49% (n=20864). Only 17% of household heads in the federal capital (n=488) were reported to have no education. A variety of occupations were reported for the main household breadwinner. In 61% of households nationally (n=56925) the main breadwinner was unemployed or in unskilled labour (this also included the few said to be students, retired or housewife/husband). The proportion of household breadwinners unemployed or working as unskilled labour varied from 57% in Punjab (n=20772) to

2 15019/24565 male respondents had any education compared with 8259/32601 female respondents. OR 4.88, 95% CI 4.70-5.07, stratified by province and urban/rural location

Household information base • 57,321 households • 433,107 people • 96,010 adult men • 87,619 adult women • 49,559 boys aged 5-12 years • 43,242 girls aged 5-12 years

Table 1. The household sample by province Province No.

HH HH size

Total people

Sindh 10,800 7.34 80,503 Balochistan 12,244 8.42 103,110 NWFP 12,901 7.97 103,799 Punjab 20,887 6.80 142,864 Islamabad 489 5.79 2,831

The mean HH size is shown weighted

Table 2. Reported male and female household members by province

Province Adults 5-12 yrs Sindh M 17,372 9,480 F 15,858 7,928 Balochistan M 21,327 12,791 F 20,698 11,544 NWFP M 24,007 11,754 F 21,091 10,261 Punjab M 32,307 15,448 F 29,047 13,529 Islamabad M 992 185 F 924 178

Other information sources (2002) • 380 community profiles • 751 government schools • 310 government health facilities • 757 union nazims and councillors • 374 male focus group • 365 female focus groups

23

76% in Balochistan (n=12138), with just 38% in this category in Islamabad households (n=487). Household construction and occupancy Nationally, just under half (48%, n=57012) the households in the survey were recorded as having a poor roof construction: katchi (mud or bamboo thatch), a rough shelter, or wooden planks. The proportion of households with poor roof construction was highest in Balochistan at 84% (n=12221) and lowest in Punjab at 41% (n=20733). Just 13% of the households surveyed in Islamabad district had a poor roof construction (n=484). Overall room occupancy was calculated by dividing the number of household members by the number of rooms in the household. Households were defined as crowded when the calculated value for room occupancy was more than 4.0. Some 30% (n=56914) of households nationally were crowded according to this definition, ranging from 27% in Punjab (n=20728) to 40% in Sindh (n=10748), and being just 6% (n=480) in Islamabad district. Vulnerable households A combined vulnerability index was constructed from the information about type of roof construction, room occupancy and type of occupation of the main household breadwinner. A household was defined as vulnerable if it had any two of the three factors: poor roof construction, overcrowding or poor occupation of the main breadwinner. If all three factors were present, a household was defined as very vulnerable. Nationally, nearly half the households (46%, n=56251) were in the ‘vulnerable’ category and 14% were in the ‘very vulnerable’ category. The highest proportion of ‘vulnerable’ households was in Balochistan (74%, n=12048) and the lowest in Punjab (38%, 20470). The proportion of ‘very vulnerable’ households was quite similar in Sindh (19%, n=10560) and Balochistan (23%, n=12048), and lowest in Punjab (10%, n=20470). There were few vulnerable (10%) or very vulnerable (1%) households in the federal capital district. In the remainder of the analysis in this report, we have used this vulnerability categorization of households to allow us to examine the situation of the most disadvantaged section of the population in comparison with the rest of the population.

A house with poor roof construction Photo: CIET

Vulnerable households A household is “vulnerable” if it has any two of: >Poor roof construction >High room occupancy >Poor occupation of main breadwinner

Very vulnerable households A household is “very vulnerable” if it has: >Poor roof construction and >High room occupancy and >Poor occupation of main breadwinner

24

Overall satisfaction with public services Household respondents were asked about their overall satisfaction with a range of public services, rating their satisfaction on a five-point scale between “very satisfied” and “very dissatisfied”. In all the following sections satisfied includes those who responded “very satisfied” or “satisfied” and not satisfied includes those who were neutral or specifically dissatisfied. Some respondents gave the answer that the service in question was not available to them at all. Note that the findings described here are the expressed perceptions of the household respondents. As described below, in many cases these can be readily understood in relation to the objective state of services in different parts of the country. In addition, where possible we have tried to relate them to the objective state of services in each site. However, when comparing “satisfied” as reported by a household respondent in a remote village in Balochistan with “satisfied” as reported by a resident of an urban site in, say, Lahore or Karachi, it should be remembered that their level of expectation about services might be different. As the first part of the discussion in the gender stratified focus groups in the survey communities, the participants talked about “the most important public service problems they would like to tell the union or zila nazim or officials about”. As expected, a wide range of problems was mentioned. Areas that featured most commonly included: water problems, problems with health facilities and lack of health facilities, problems with roads and transport, and various problems with schools or lack of functioning schools. People were concerned about the basic necessities and basic support services.

Roads and transport Nationally, about a third (31%, n=46706) of households said they were satisfied with the roads in their area, 51% were not satisfied and 18% reported they had no roads at all. The perceptions were worst in Balochistan, where only 17% (n=10103) were satisfied with roads and best in NWFP, where 40% (n=10744) were satisfied (Figure 1) In focus group discussions in Balochistan, the problems of roads and transport were mentioned but rather less often than in focus groups in the other provinces. It could be that focus group participants in Balochistan felt there was little

31

67

32

40

17

24

0 20 40 60

Pakistan

Isb

Punjab

NWFP

B'stan

Sindh

Fig 1. % households satisfied with roads

“We have no gas, no electricity, no employment. We have no facilities. So what services are you talking about?” Female focus group, Rajan Pur

25

point in raising this intractable problem with the union and district governments and were prioritizing problems for which there might be a more immediate solution. Household respondents living in urban sites were more likely to be satisfied with the available roads, taking into account province, compared with those living in rural sites3. This was especially marked in Balochistan4. Overall, respondents from ‘very vulnerable’ households were marginally less likely to be satisfied with the available roads with those from less vulnerable households5. The association was mainly in Balochistan6 and particularly in urban sites of Balochistan7. Thus, especially in Balochistan, sites with a high proportion of ‘very vulnerable’ households were particularly poorly served by roads. The lack of adequate roads also contributes to the poverty of these communities, for example by limiting their access to markets. Across the country, 59% of households (n=46304) were satisfied with the public transport in their area, 30% were not satisfied, and 11% said they had no public transport at all. Satisfaction with public transport was quite similar across the provinces, though somewhat lower in Balochistan (Figure 2). Urban dwellers were about twice as likely to be satisfied with public transport compared with rural dwellers8. Respondents from ‘very vulnerable’ households were less likely to be satisfied with public transport compared with those from less vulnerable households9. Male household respondents were somewhat less likely to be satisfied with public transport compared with female respondents10, perhaps because men use the services more than women.

34033/12391 urban dwellers were satisfied compared with 9388/34315 rural dwellers. OR 1.33, 95% CI 1.27-1.39, after stratification by province. 4 In Balochistan 788/2549 urban dwellers were satisfied compared with 870/7554 rural dwellers. OR 3.44, 95% CI 3.07-3.85 5 1867/7547 from very vulnerable households satisfied compared with 11312/38322 from less vulnerable households. OR 0.93, 95% CI 0.88-0.99, stratified by province, urban/rural 6 In Balochistan, 245/2224 from very vulnerable households satisfied with roads compared with 1379/7715 from less vulnerable households. OR 0.66, 95% CI 0.57-0.77 stratified by province, urban/rural. 7 In urban sites of Balochistan, 61/354 respondents from very vulnerable households are satisfied with roads compared with 708/2153 from less vulnerable households. OR 0.42, 95% CI 0.31-0.57 8 7923/12256 urban dwellers are satisfied with public transport compared with 16582/34048 rural dwellers. OR 1.97, 95% CI 1.88-2.06, after stratification by province. 9 3302/7473 from very vulnerable households are satisfied with public transport compared with 20748/38001 from less vulnerable households. OR 0.73, 95% CI 0.70-0.77, after stratification by province. 10 10897/20749 male respondents were satisfied with public transport compared with 13604/25536 female respondents. OR 0.83, 95% CI 0.77-0.88, after stratification by province, urban/rural

59

52

62

57

36

55

0 20 40 60

Pakistan

Isb

Punjab

NWFP

B'stan

Sindh

Fig 2. % households satisfied with public transport

“If the government had been able to solve our problems we would not be living in this jungle.” Male focus group, Sanghar

26

Sewage/sanitation Nationally, just 12% (n=46689) of households said they were satisfied with the sewage/sanitation service in their area, 37% were not satisfied and over half (51%) considered they had no government sewage service at all. The perceptions about government sewage services were negative in all provinces, and worst in Balochistan, where only 5% (n=10091) were satisfied with the service (Figures 3 and 4). Male household respondents were somewhat less likely to be satisfied with government sewage services compared with female respondents11. When specifically asked about the sewage system, household respondents living in urban sites were more than four times as likely to be satisfied with government sewage services compared with those living in rural sites12. The contrast between urban and rural sites was particularly marked in Sindh and Balochistan. Sewage systems problems were quite often mentioned as a priority problem in focus groups. Interestingly, however, focus groups in urban sites mentioned sewage as a priority problem to be raised with the union and district government rather more often than groups in rural sites, perhaps because they have a system that needs improvement whereas in rural sites there is no system and none expected. Respondents from ‘very vulnerable’ households were less likely to report satisfaction with the government sewage service, taking into account province, compared with those from less vulnerable households13. This relationship was found in both urban and rural locations. The views of households about the sewage services available to them were related to the sewage services observed in the communities (recorded in the community profile for each community – see Annex 7). Households in sites where there was some form of government sewage service in place were indeed more likely to say they had

11 1968/20921 male respondents were satisfied with sewage services compared with 2919/25749 female respondents. OR 0.71, 95% CI 0.67-0.77, after stratification by province, rural/urban 122595/12396 urban dwellers were satisfied with sewage services compared with 2292/34293 rural dwellers. OR 4.05, 95% CI 3.80-4.32, after stratification by province. 13 425/7541 very vulnerable households were satisfied with sewage services compared with 4345/38309 less vulnerable households. OR 0.73, 95% CI 0.65-0.81, after stratification by province, urban/rural

12

36

14

15

5

11

0 10 20 30 40

Pakistan

Isb

Punjab

NWFP

B'stan

Sindh

CIET2003

Fig 4. % households satisfied with government sewage services

Fig 3. % households satisfied with sewage services

27

access to a government sewage system14. ‘Very vulnerable’ households were less likely to be located in communities where there was a government sewage system in place compared with less vulnerable households15: sewage systems were more likely to be provided for better off communities. The presence of a system in a community did not necessarily mean that all households have access. Even within sites where there was a government sewage system in place, ‘very vulnerable’ households were less likely to say they had access compared with less vulnerable households16. Households in communities where there was a “good” government sewage system in place (all covered/piped or partly open in pukki drains) were more likely to be satisfied with the government sewage system compared with households in communities where there was no government system or only a poor system17. ‘Very vulnerable’ households were less likely to be in sites with a good government system in place, compared with less vulnerable households18. Even within sites with a good system in place, ‘very vulnerable’ households were less likely to be satisfied, compared with less vulnerable households19. This may be because they did not themselves benefit from the system.

Garbage disposal Nationally, few people were satisfied with the government garbage disposal service available to them. Only 6% (n=46611) of households were satisfied with government garbage disposal services, 22% were dissatisfied and 71% said they had no such service available. Satisfaction was low across all provinces (Figure 5). In the federal capital district, rather more people were satisfied with government garbage disposal services (31%, n=488).

14 11736/14719 households in communities where there was a government sewage system in place said they had access to a system compared with 7606/31353 in communities without a government system. OR 6.31, 95% CI 5.99-6.66, stratified by province and urban/rural 15 1403/7485 very vulnerable households were in sites with a government sewage system compared with 13023/24878 less vulnerable households. OR 0.76, 95% CI 0.71-0.81, stratified by province, urban/rural 16 In sites with a government sewage system in place, 959/1399 very vulnerable households said they had access compared with 10488/12982 less vulnerable households. OR 0.52, 95% CI 0.46-0.59 17 1791/7269 households in sites with a good government sewage system were satisfied compared with 2927/38677 in sites without a good system. OR 2.00, 95% CI 1.85-2.16, stratified by province, urban/rural 18 509/7476 very vulnerable households are in sites with a good government sewage system compared with 6596/37786 less vulnerable households. OR 0.63, 95% CI 0.57-0.70, stratified by province, urban/rural 19 In sites with a good government sewage system, 96/506 very vulnerable households were satisfied compared with 1641/6578 less vulnerable households. OR 0.70, 95% CI 0.56-0.89

6

31

7

7

4

10

0 10 20 30 40

Pakistan

Isb

Punjab

NWFP

B'stan

Sindh

Fig 5. % households satisfied with government garbage disposal service

Very vulnerable households had less access to government sewage systems, even in communities where such systems were in place.

28

Government garbage disposal was more or less limited to urban settings, so it is not surprising that, when specifically asked, households in urban sites were much more likely to be satisfied with the garbage disposal service compared with households in rural sites20. Many more focus groups in urban sites, compared with rural sites, prioritized garbage disposal as a problem to take to the union and district government, reflecting greater concern about garbage as an issue in urban sites. The build up of garbage is a much more visible problem in urban settings, whereas garbage can more easily be dumped away from the household in rural settings. A respondent from a ‘very vulnerable’ household was less likely to be satisfied with government garbage disposal services compared with a respondent from a less vulnerable household21. This was true in both urban and rural sites. The garbage disposal arrangements in the survey sites were noted by the field teams (see Annex 7). Household perceptions about government garbage disposal services were related to the systems in place in the communities. As expected, urban communities were much more likely to have a government garbage disposal system compared with rural communities22. Households in sites where there was a government garbage disposal system in place were more likely to report access to a system compared with households with no system in place23. Those households reporting access in places where there was no government system might have been referring to non-government systems. ‘Very vulnerable’ households were less likely to be in sites with a government garbage system compared with less vulnerable households24. And even in sites with a government system in place, vulnerable households were less likely to report access compared with less vulnerable households25.

20 2357/12371 urban households were satisfied with garbage disposal compared with 366/34240 rural households. OR 21.32, 95% CI 18.97-23.95, after stratification by province. 21 184/7527 very vulnerable households were satisfied with garbage disposal compared with 2471/38251 less vulnerable households. OR 0.65, 95% CI 0.56-0.77, after stratification by province, urban/rural 22 6876/12065 urban households are in sites with a government garbage disposal system compared with 1244/34293 rural households. OR 29.29, 95% CI 27.40-31.31, stratified by province 23 6402/8064 households in sites with a government garbage disposal system reported access to a system compared with 6033/38057 in sites without a government system. OR 5.96, 95% CI 5.53-6.41, stratified by province and urban/rural 24 671/7508 very vulnerable households are in sites with a government garbage disposal system compared with 7260/33003 less vulnerable households. OR 0.69, 95% CI 0.62-0.76, stratified by province, urban/rural 25 In sites with government garbage disposal, 453/668 very vulnerable households reported access compared with 5797/7215 less vulnerable households. OR 0.52, 95% CI 0.43-0.62

Very vulnerable households had less access to government garbage disposal systems, even in communities where such systems were in place.

29

Gas supply Across the country, just 19% (n=46578) of households were satisfied with the government gas supply, 2% were dissatisfied with the service and 79% said they had no such service. It seems that if a household had a gas supply at all, they were satisfied with the service. The variation in household satisfaction with gas supply by province is shown in Figure 6. The provincial satisfaction was highest in Sindh and lowest in NWFP. Household respondents living in urban sites were much more likely to be satisfied with the government gas supply compared with those living in rural sites26. This was mainly because only around 5% of households in rural sites reported having a government gas supply, compared with about 60% in urban sites27. Taking province and urban/rural location into account, there was no difference between male and female household respondents in the proportion expressing satisfaction with gas supply. However, many more female focus groups, compared with male groups, prioritized gas supply as a problem to take to the union or district government. Women probably felt the lack of a gas supply for cooking more than men did. Respondents from ‘very vulnerable’ households were less likely to report satisfaction with the government gas supply compared with those from less vulnerable households28. This was partly explained by the higher proportion of ‘very vulnerable’ households in rural settings, but within urban locations ‘very vulnerable’ households were less likely to be satisfied with the government gas supply29 (or to have a gas supply)30. In effect, ‘very vulnerable’ households in urban settings were excluded from the gas supply. Since the supply is usually to a whole area, it indicates that the poorer areas of towns and cities were not prioritized for provision of a gas supply.

26 6784/12374 urban dwellers were satisfied compared with 1076/34204 rural dwellers. OR 36.61, 95% CI 34.00-39.43, after stratification by province. 27 7497/12374 urban dwellers have a gas supply compared with 1783/34204 rural dwellers. OR 26.98, 95% CI 25.32-28.75, after stratification by province. 28 469/7516 very vulnerable households were satisfied with the gas supply compared with 7195/38226 less vulnerable households. OR 0.43, 95% CI 0.38-0.48, after stratification by province, urban/rural 29 In urban sites, 287/1039 very vulnerable households are satisfied with gas supply compared with 6320/11030 less vulnerable households. OR 0.28, 95% CI 0.25-0.33 30 In urban sites, 353/1039 very vulnerable households have a gas supply compared with 6945/11030 less vulnerable households. OR 0.30, 95% CI 0.26-0.35

19

46

19

11

19

39

0 20 40

Pakistan

Isb

Punjab

NWFP

B'stan

Sindh

Fig 6. % households satisfied with government gas supply

30

Electricity supply Satisfaction with the electricity supply was relatively high, with nearly two thirds of households (63%, n=46634) reporting they were satisfied with the electricity supply in their area. Some 21% of households were not satisfied with the electricity supply, and 16% said there was no electricity supply in their area. Satisfaction with the electricity supply was lowest in Balochistan (Figure 7). Balochistan also had the highest proportion of households (32%) that reported they had no electricity supply. Households in urban sites were more than three times more likely to be satisfied with the electricity supply compared with households in rural sites31. Similarly, households in urban sites were much more likely to have any electricity supply compared with households in rural areas32. Male household respondents were less likely to express satisfaction with the electricity supply compared with female respondents33. ‘Very vulnerable’ households were less likely to be satisfied with the electricity supply compared with less vulnerable households34, in both urban and rural locations. ‘Very vulnerable’ households were less likely to have any electricity supply compared with less vulnerable households35. Again, this was true in both urban and rural locations.

31 9552/12371 urban dwellers were satisfied with the electricity supply compared with 17132/34263 rural dwellers. OR 3.35, 95% CI 3.20-3.52, after stratification by province 32 12014/12371 urban households had an electricity supply compared with 25461/34263 rural households. OR 12.59, 95% CI 11.30-14.04, after stratification by province. 33 11146/20908 male respondents were satisfied with the electricity supply compared with 15528/25707 female respondents. OR 0.67, 95% CI 0.65-0.70, after stratification by province, urban/rural 34 3325/7535 very vulnerable households were satisfied with electricity supply compared with 22841/38262 less vulnerable households. OR 0.72, 95% CI 0.68-0.76, after stratification by province, urban/rural 35 5124/7535 very vulnerable households have any electricity supply compared with 31643/38262 less vulnerable households. OR 0.64, 95% CI 0.60-0.68, after stratification by province, urban/rural

63

89

72

56

48

55

0 50

Pakistan

Isb

Punjab

NWFP

B'stan

Sindh

Fig 7. % households satisfied with electricity supply

31

Use and experience of government health services

Household satisfaction with government health services About a quarter (23%, n=46396) of households across the country said they were satisfied with the government health services in their area, nearly half (45%) were dissatisfied with the available service, and 32% considered they had no government health service at all. The general household satisfaction with government health services was fairly similar across all provinces (Figure 8), with satisfaction lowest in Balochistan (17%). Note that this satisfaction rating came from all households, whether or not they used government health services. Dissatisfaction with government health services – perhaps coloured by media or personal reports of the experience of others – is one reason for not choosing to use these services. Male household respondents were somewhat less likely to report satisfaction with government health services compared with female respondents36. Households in urban sites were more likely to be satisfied with the government health services in their area compared with households in rural sites37. Similarly, households in urban sites were more likely to report there was a government health facility available to them38. The difference in perceived availability of government health services between urban and rural settings was mainly a feature of Sindh39 and NWFP40 provinces. Households in sites with a government health facility within 5km (information from the community profile, see Annex 7) were more than four times as likely to report they had a government health service available compared with households in sites without a facility within 5km41.

36 4300/20803 male respondents were satisfied with government health services compared with 6127/25574 female respondents. OR 0.79, 95% CI 0.75-0.82, after stratification by province, urban/rural 37 3534/12248 urban dwellers were satisfied compared with 6895/34148 rural dwellers. OR 1.65, 95% CI 1.57-1.73, after stratification by province 38 9071/12248 urban households report availability of government health services compared with 23685/34148 rural households. OR 1.53, 95% CI 1.46-1.61, after stratification by province 39 In Sindh, 2079/3235 urban households report available government health services compared with 1776/5269 rural households. OR 3.54, 95% CI 3.22-3.89 40 In NWFP, 1260/1437 urban households report available government health services compared with 6903/9251 rural households. OR 2.42, 95% CI 2.04-2.87 41 26971/34582 households in sites with a government facility within 5km reported a service available compared with 5415/11324 in other sites. OR 4.36, 95% CI 4.16-4.58, stratified by province, urban/rural

23

31

23

27

17

22

0 10 20 30 40

Pakistan

Isb

Punjab

NWFP

B'stan

Sindh

Fig 8. % households satisfied with government health services

“May God save even our enemies from being in hospitals” Female focus group, Abbotabad.

32

Respondents from ‘very vulnerable’ households were less likely to be satisfied with government health services compared with those from less vulnerable households42. Respondents from ‘very vulnerable’ households were also less likely to report they had a government health service available to them, compared with less vulnerable households43. The difference in availability of government health services to ‘very vulnerable’ and less vulnerable households was more marked in rural areas. Indeed, ‘very vulnerable’ households were less likely to be within 5 km of a government health facility (information from community profile)44. It seems that poor communities with a high proportion of ‘very vulnerable’ households were poorly served by government health facilities, although they were probably in more need of them than their better off neighbours.

Access to health care Household respondents were asked where people in the household usually went for medical attention. Nationally, nearly a third of households (31%, n=57075) reported they usually used government health facilities, while nearly half (47%) reported they usually used private qualified practitioners, and 21% usually used private unqualified practitioners. A small number used NGO or services facilities. The utilization pattern varied from one province to another (Figure 9). In NWFP a relatively high proportion of households reported using government health facilities (59%, n=12850). In Sindh more than two thirds of households reported using private qualified practitioners (68%, n=10694). In Punjab more than a third of households (34%, n=20832) reported they usually used non-qualified private practitioners for medical attention. Urban households overall were less likely to report using government health facilities compared with rural households45.

42 1398/7501 very vulnerable households are satisfied with government health services compared with 8826/38062 less vulnerable households. OR 0.86, 95% CI 0.81-0.92, stratified by province, urban/rural 43 4739/7501 very vulnerable households reported available government health service compared with 27430/38062 less vulnerable households. OR 0.79, 95% CI 0.75-0.84, stratified by province, urban/rural 44 5591/7508 very vulnerable households are in sites within 5km of a government health facility compared with 28719/38003 less vulnerable households. OR 0.87, 95% CI 0.82-0.92, stratified by province, urban 45 4297/15463 urban households use government health facilities compared with 17842/41612 rural households. OR 0.60, 95% CI 0.58-0.63, after stratification by province.

Fig 9. % households who usually use government health facilities

31

47

24

59

50

23

0 20 40 60

Pakistan

Isb

Punjab

NWFP

B'stan

Sindh

“The health staff discriminate – beautiful and well off people get better services.” Female focus group, Jampur

33

Overall ‘very vulnerable’ households were not more likely to make use of government health services compared with less vulnerable households46. However, in Sindh ‘very vulnerable’ households were marginally more likely to use government health facilities compared with less vulnerable households47. In urban sites across the country, ‘very vulnerable’ households were 50% more likely to use government health services compared with less vulnerable households48. In urban sites in Sindh (a big part of this section is accounted for by Karachi and Hyderabad), ‘very vulnerable’ households were 70% more likely to use government health services compared with less vulnerable households49. In Karachi, especially, the less vulnerable households had a good choice of alternative private health services. The picture that emerged is that households with the opportunity to use private health services often chose the private option. Those who could not afford to use private services, the ‘very vulnerable’, had no choice but to use the government services. The contrast was especially marked in urban Sindh, where there is a greater availability of private health care, and in urban sites in general, where there are more private health services available. As shown in the map (Figure 10), a lower proportion of households in Sindh and much of Punjab used government health facilities than elsewhere in the country. As part of the community profile, we noted the distance of health facilities from the sample communities (see Annex 7). Distance made a difference to preferred health care providers. Households in communities with a government health facility within 5 km were nearly one and a half times more likely to use government health facilities compared with those in communities further from government facilities50. This was true whether or not there was also a private facility within 5 km.

46 3822/8956 very vulnerable households use government facilities compared with 17963/47063 less vulnerable households. OR 1.03, 95% CI 0.98-1.08, after stratification by province, urban/rural 47 In Sindh, 710/2299 very vulnerable households use government facilities compared with 1939/8161 less vulnerable households. OR 1.10, 95% CI 0.99-1.23, stratified by urban/rural 48 In urban sites, 450/1209 very vulnerable households use government facilities compared with 3756/13836 less vulnerable households. OR 1.59, 95% CI 1.40-1.80 49 In Sindh urban sites, 54/211 very vulnerable households use government health services compared with 660/3952 less vulnerable households. OR 1.72, 95% CI 1.23-2.40 50 14805/34770 households in sites within 5 km of a government health facility use government health facilities compared with 3450/11375 in sites further from facilities. OR 1.46, 95% CI 1.39-1.53, stratified by province, urban/rural

Fig 10. Proportion of households who usually use government health services

CIET2003

“With money in hand one can get all facilities.” Female focus group, Abbotabad (urban site)

34

Views from community focus groups In gender stratified focus groups in the survey communities, participants were asked about “what changes in government health services would make people feel more satisfied and more likely to use the services”. Virtually all the focus groups mentioned that the government health services needed more staff and better qualified staff. And most groups simply said that more services and facilities were needed so as to improve access. Another commonly voiced requirement was for “more medicines” so that the need to buy the required medicines outside the health facility would be reduced. A fair number of groups described the need to improve the behaviour of staff in government health services, citing examples of bad staff behaviour. Some groups mentioned not only changes as they would be felt by the service users, but also systems changes to achieve these ends. So, for example, about a third of focus groups mentioned the need for strict monitoring of the service providers, with ways of punishing those who were not doing their job properly. A minority of groups mentioned the need to prevent corruption and political interference in government health services, citing the problems they believed were being caused by political interference. Travel to health facilities Four out of ten households (42%, n=56622) reported they incurred some travel cost to reach the health service or practitioner they usually used for medical attention, with a mean cost among those who paid anything of Rs 71. The proportion paying travel costs was highest among households using private qualified practitioners or NGO/service facilities and lowest among households using unqualified practitioners (Table 3). Among those who paid anything, travel costs to government facilities were highest, and lowest among those using unqualified practitioners. Over half of households (55%, n=56792) reported they walked to their usual health service or practitioner, while about a third (37%) used public transport and 8% hired transport. Effectively, those who usually walked to the health facility were those who did not pay any travel costs.

Table 3 Costs of travel to get health care Service % who

paid Mean cost

(Rs.)* All services 42 71 Government 45 89 Private qualified 51 66 Unqualified 16 29 NGO/services 52 59

*Among those who paid anything for travel

“The government does nothing. The doctors treat humans like donkeys.” Female focus group, Karak

“We get fourth grade medicines.” Male focus group, Haripur

35

Some of the apparent higher travel costs to reach government health services (Table 3) were because more people in NWFP and Balochistan used government health services, and travel costs were higher in these provinces due to the difficult terrain and longer distances. Among those households who usually used government health facilities, the proportion having to pay travel costs to reach government health facilities was highest in NWFP (Table 4). Mean costs for travel among those who paid anything were higher for Balochistan and NWFP. ‘Very vulnerable’ households who usually used government health services were less likely to pay travel costs, taking province into account51. However, this was not true in all places and in rural Sindh, ‘very vulnerable’ households were actually 50% more likely to incur travel costs to government health facilities52. Whenever possible, household experiences about government health services were linked to features of the particular facility they visited, assessed as part of the data collection exercise in the communities. Households whose usual facility was within the community were much less likely to report travel costs for visits in the last three months53.

Payments for health services Table 5 shows the costs of different elements for visits to a government health facility in the last three months, for a case of fever. This was the commonest reason for visiting a health facility. Restricting the costs analysis to a particular type of case means that variations in costs due to variations between different diagnoses and types of treatment are reduced. Overall, some 93% (n=6731) of the service users reported they paid something for the visit. Service users were asked for information about the costs of separate times, but some of them could not give this detail and for them only the total cost was recorded. The mean total cost for a visit to a government facility with fever, whether or not separate elements could be reported, was Rs 232. This cost was exceptionally higher in Islamabad (Rs 963) (see Annex 6) but this was probably linked to a different policy for service charges at the government health facilities in the

51 1536/3797 very vulnerable households paid for travel to government health facilities compared with 8350/17877 less vulnerable households. OR 0.83, 95% CI 0.77-0.90, stratified by province, urban/rural 52 In rural Sindh, 290/652 very vulnerable households paid for travel to government health facilities compared with 447/1276 less vulnerable households. OR 1.49, 95% CI 1.22-1.81 53 4176/8614 whose health facility was in the community paid travel costs compared with 333/3235 users of facilities outside the site. OR 8.22, 95% CI 7.21-9.37, stratified by province, urban, road condition

Table 5. Costs of visits to government health facilities for a case of fever S B N P Pak Total cost % who paid 87 81 99 97 93 Mean cost 175 303 347 184 232 Ticket % who paid 29 18 87 90 68 Mean cost 6 3 4 3 4 Meds in % who paid 3 2 6 9 6 Mean cost 77 165 129 154 198 Meds out % who paid 73 79 86 47 61 Mean cost 148 292 292 295 253 Providers % who paid 7 4 4 4 5 Mean cost 47 116 143 128 93 Other costs % who paid 1 2 6 2 2 Mean cost 107 256 184 102 159 *Costs are in Rs, among those who paid S=Sindh, B=Balochistan, N=NWFP, P=Punjab, Pak=Pakistan

Table 4 Costs of travel to government health facilities

Area % who paid

Mean cost (Rs.)*

Sindh 38 62 Balochistan 34 185 NWFP 62 116 Punjab 43 54 Islamabad 60 85 Pakistan 45 89

*Among those who paid anything for travel

36

federal capital. Among the four provinces NWFP had the highest mean total costs for visits to the health facility (Rs 347). Most common payments made were for outpatient registration ticket (68%, n=5189) or for medicines from outside the health facility (61%, n=5161). Only 5% said they made a payment directly to a service provider, in an average amount of Rs 93. The average costs of a visit to a government health facility with fever in the previous three months can be compared with the costs of a visit to a private qualified practitioner with fever in the same period. These are shown in Table 6. Overall, the mean total cost of a visit to a government health facility for a case of fever (Rs 232) was not much less than the cost of a visit to a private qualified practitioner for a case of fever (Rs 250). For users of government health services with fever in the last three months, there was no important difference in making a payment overall between male and female service users. Nor were service users from ‘vulnerable’ or ‘very vulnerable’ households less likely to pay than those from less vulnerable households. Similarly, among those who paid, the amounts were not different by sex of the service users, nor by vulnerability category of the service user’s household. Overall, there was no clear picture of urban dwellers being more or less likely to pay than rural dwellers: they were more likely to pay in some provinces and less likely to pay in others. The amount paid was not significantly different between urban and rural dwellers.

Satisfaction with health service contacts Households were asked about their satisfaction with health services in relation to visits to the services in the last three months. • Six out of every ten service users (62%, n=18707) were

satisfied with their contact with government health services.

• Eight out of ten service users (82%, n=21875) were satisfied with their contact with private qualified practitioners.

• Three quarters (76%, n=7954) were satisfied with their contact with unqualified practitioners.

The satisfaction with government health service visits was lowest in Balochistan (48%, n=5426) (Figures 11 and 12)

Table 6. Costs of visits to private qualified practitioner for a case of fever S B N P Pak Total cost % who paid 98 93 99 99 98 Mean cost 194 618 386 237 250 Ticket % who paid 11 2 16 10 9 Mean cost 30 34 48 40 39 Meds in % who paid 16 8 22 49 33 Mean cost 70 192 230 103 103 Meds out % who paid 81 82 76 34 59 Mean cost 177 397 386 289 237 Providers % who paid 60 59 47 43 51 Mean cost 47 134 86 104 79 Other costs % who paid 2 2 5 2 2 Mean cost 113 389 173 565 399 *Costs are in Rs, among those who paid S=Sindh, B=Balochistan, N=NWFP, P=Punjab, Pak=Pakistan

62

70

64

60

48

70

0 20 40 60 80

Pakistan

Isb

Punjab

NWFP

B'stan

Sindh

Fig 11. % service users satisfied with visits to government health facilities

37

People were less satisfied with their contacts with government health services in areas where they were using them more. In Balochistan and NWFP, use of government health services was relatively high (Figure 10) but satisfaction with contacts with these services was relatively low (Figure 12). This suggests that people did not choose to use government health services on the basis of satisfaction with previous contacts, but because they had no other choice in their area, or they could not afford to pay for private services. The most common reason given for satisfaction with the services among those who used government health facilities was “good treatment” (80%; 10245/12779), as the health problem got better. The other main reasons for satisfaction were good staff at the health facility (15%; 1909/12779) and good facilities or services (2%; 245/12779). For those service users who were dissatisfied with the government health services, the most common reason was that the treatment was bad (58%; 5452/9097 as their health problem was not solved. Another common reason for dissatisfaction was non-availability of medicines inside the facility (14%; 1289). Poor facilities or services (7%; 637), bad staff (5%; 557), too costly (3%; 310) and doctors being not available (3%; 247) were some other less frequent reasons mentioned by those who were dissatisfied. There was no difference in satisfaction of government health service users by sex of the household respondent, nor by sex of the service user (including when only adult services users were considered). A small difference between urban and rural service users was explained when province and household vulnerability were taken into account. Service users from ‘very vulnerable’ households were less likely to be satisfied with the service compared with service users from less vulnerable households54. Some 42% of the government health facilities serving the sample communities said they made efforts to let the community know about their services. People who used a health facility that made efforts to publicise their services were more likely to be satisfied compared with people whose health facility was not active in this way55.

54 1771/3336 service users from very vulnerable households were satisfied with government health services compared with 8839/15083 from less vulnerable households. OR 0.83, 95% CI 0.77-0.90, stratified by province and urban/rural location 55 2549/4307 people using a health facility that informs the public about services were satisfied compared with 3469/6163 using a facility that does not inform the public. OR 1.23, 95% CI 1.13-1.33, stratified by province, urban/rural, sex of respondents, vulnerability

CIET2003

Fig 12. Proportion service users satisfied with visits to government health facilities

Service users were more likely to be satisfied with health facilities that were trying to keep the public informed about their services.

“The doctors sell government medicines at their private stores.” Male focus group, Gwadar

“Doctors got their degrees through unfair means and don’t know what to prescribe for patients.” Male focus group, Hyderabad

38

Monitoring government health services Making complaints Only 10% of households (n=21840) who usually used government health services knew how to make a complaint about the service if they were not satisfied with any aspect. The proportion of households knowing how to complain about government health services was low in all provinces, and somewhat higher in the federal capital (26%) (Figure 13). Male household respondents were more than four times more likely to know how to make a complaint about government health services compared with female respondents56. This was true in all provinces, but was particularly marked in Balochistan57. Respondents from households categorized as ‘very vulnerable’ were less likely to know how to complain about government health services compared with those from less vulnerable households58. The services themselves can make a difference. People whose usual health facility had a system for complaints (45% of the reviewed facilities, see Annex 8) were 30% more likely to know how to complain59, taking other factors into account. Monitoring by citizens Male and female focus groups in each of the survey communities heard about the satisfaction of households and service users with government health services in their own district. They were then asked about how Citizen Community Boards (CCBs) could monitor government health services in order to improve their service to the public. About 10% of the focus groups (evenly split between male and female groups) were not able to formulate any ideas

56 1686/9848 male respondents knew how to complain about government health services compared with 533/11979 female respondents. OR 4.48, 95% CI 4.04-4.97, after stratification by province, urban/rural 57 In Balochistan, 463/1747male respondents knew how to complain about government health services compared with 109/4400 female respondents. OR 14.20, 95% CI 11.35-17.80 58 188/3780 respondents from very vulnerable households knew how to complain about government health services compared with 1981/17712 from less vulnerable households. OR 0.46, 95% CI 0.39-0.54, stratified by province, urban/rural 59 769/6705 people who used a service with a complaints system knew how to complain compared with 442/5297 whose service had no complaints system. OR 1.29, 95% CI 1.12-1.48, stratified by province, urban/rural, sex of respondent and vulnerability

10

27

10

11

8

13

0 10 20 30

Pakistan

Isb

Punjab

NWFP

B'stan

Sindh

Fig 13. % households who knew how to complain about government health services

“Those who don’t work should be fired so the others learn the lesson.” Female focus group, Khanewal

39

about how citizens could monitor government health services. The rest of the groups commonly mentioned that they would want to monitor staff attendance, punctuality and behaviour, as these were common complaints from patients. Many groups also mentioned the need to monitor and check medicines and supplies in government health facilities, reflecting a concern that medicines were not available because they were diverted from government health facilities onto the private market. Male groups were more likely to want staff monitoring and punishment for bad behaviour as a means of improving government health services. Some participants of female focus groups mentioned there would be risks for ordinary citizens trying to monitor government health services. In some focus groups, participants gave quite specific details about what information they would need if they were to monitor government health services, for example through a CCB. For example, they mentioned they would need the records of medicines supply and distribution, records of intended and actual timings of opening hours and clinics, and details of staffing of facilities. The focus groups also offered some interesting suggestions for ways in which, if they were CCB members, they could go about collecting the information they needed for monitoring government health services. About two thirds of the groups mentioned making surprise visits to health facilities or posing as patients to see how the staff behaved. Interestingly, this ‘direct’ monitoring approach was more popular in female focus groups. Another quite common suggestion was to ‘ask patients or community members’. Thus the idea of going out and seeking information from the population about their experience of services, in order to monitor those services, seemed quite natural to many people. Collecting data and using it for monitoring, as part of the activities of CCBs, would not be a strange concept for many ordinary people. When asked about how they might report on any monitoring of health services to district officials, almost all groups mentioned a personal visit to the officials as a first choice method. Most groups also mentioned submitting written reports to the officials. About a third of the focus groups suggested they could report to district officials via the union council nazim. A few groups felt the media should be involved in the process of reporting on the findings of monitoring health services. Male groups mentioned this as an option more often than female groups.

“If we try to monitor them they would not let us and they may even lash us out of the facility.” Female focus group, Sahiwal

“Working on branches and leaves does not make a tree healthy and green – one has to focus on its roots.” Female focus group, Lahore

40

Views of government health service providers As part of the data collection, government health facilities serving the sample communities were visited and the head of the facility or deputy was interviewed in each facility. Some 47 facilities were visited during the 2001 ten district pilot and a further 310 were visited in the 2002 main survey (see Annex 8). The instrument for interviewing service providers was refined after the 2001 pilot and the findings described here are from the 2002 main survey. About half (52%) the heads of government health facilities interviewed thought the new local government plan would improve their services and only 5% thought it would make their services worse (Table 7). Interviewees were asked about the ways in which the new plan would affect their services. The responses among those who thought services would be better or unchanged are shown in Table 8. As an insight into under-the-table payments in the government health services, the health providers interviewed were asked if they knew of anyone who had had to make a payment from their salary or to secure employment. Some 8% (21/260) admitted to knowing of such cases and 4% (11/248) admitted they themselves had experienced this. A number of interviewees declined to answer these questions, which came at the end of the interview.

Table 7. Heads of government health facilities: the local government plan

Effect on services % (n) Make them better 52 (162) No effect 32 (99) Make them worse 5 (14) Don’t know 11 (35)

Table 8. Heads of government health facilities: effects of the government plan

Response % (n) Better monitoring 24 (37) Supportive LG 21 (32) More medicines 19 (30) More funds 16 (25) Better coordination 7 (11) More staff available 7 (10) More facilities 7 (10)

41

Education services

Household satisfaction with government education services Nationally, over a half 55% (n=46194) of households said they were satisfied with government education services in their area, 38% were not satisfied, and 7% considered they had no government education services available in their area. Satisfaction with government education services was fairly evenly spread across the four provinces, being highest in NWFP at 60% (n=10642) (Figure 14). ‘Very vulnerable’ households were slightly less likely to be satisfied with government schools in their area compared with less vulnerable households60. And ‘very vulnerable’ households were less likely to say they had a government school available to them, compared with less vulnerable households61. This lower perceived availability of government schools to ‘very vulnerable’ households was found only in rural sites. Overall, the satisfaction with government education services did not differ between urban and rural sites. Households in urban communities were overall somewhat less likely to say they had a government school available compared with households in rural communities62. This differed between provinces: urban households reported slightly less government schools’ availability compared with rural sites in Punjab and Balochistan, while in NWFP and Sindh urban households reported more availability of government schools compared with rural sites. The presence of a government school within 1.5km of the community (community profile, see Annex 7) made a difference. Households within 1.5km of a government school were three times more likely to say they had access to the government education service63 and 60% more likely to say they were satisfied with government education

60 4030/7465 very vulnerable households were satisfied with government school in the area compared with 20936/37902 less vulnerable households. OR 0.94, 95% CI 0.89-0.99, stratified by province, urban/rural 61 6803/7465 very vulnerable households had government schools available compared with 35614/37902 less vulnerable households. OR 0.78, 95% CI 0.71-0.86, after stratification by province. 62 11154/12178 urban households say they have a government school available compared with 32021/34016 rural households. OR 0.82, 95% CI 0.75-0.88, stratified by province 63 40898/43518 households within 1.5km of a government school said government education was available in the area compared with 1798/2186 further from a government school. OR 3.32, 95% CI 2.94-3.77, stratified by province, urban/rural, vulnerability

55

44

53

60

50

56

0 20 40 60

Pakistan

Isb

Punjab

NWFP

B'stan

Sindh

Fig.14. % households satisfied with government education

“If we had schools our children would have been better human beings.” Female focus group, Bahawalnagar

42

services64 compared with households further from a government school.

School enrolment The net school enrolment rate of children aged 5-9 years was 67% (n=48721). The enrolment rates were higher in Punjab (70%) and NWFP (70%) compared with Sindh (62%) and Balochistan (60%). The school enrolment rate was higher (73%, n=25843) among boys than girls (60%, n=22876). A boy was twice as likely to be enrolled in school compared with a girl65. The gender gap was highest in Sindh (70% boys enrolled and 52% girls) and NWFP (78% boys enrolled and 60% girls) and lowest in Punjab (73% boys enrolled and 66% girls) .The federal capital Islamabad had the highest enrolment rate (88%) with very little gender gap (90% boys enrolled and 86% girls). The striking difference between Figures 15 and 16 illustrates the gender gap in primary education and the way this varied across the country. Disadvantaged children were less likely to attend school. A child from a ‘very vulnerable’ household was less likely to be enrolled in school compared with a child from a less vulnerable household66. As well as being found across all provinces, this relationship between vulnerability of the household and school enrolment was true for both boys and girls and in both urban and rural sites. Girls who lived in a community which had a government girls’ school within 1.5km were nearly twice as likely to be enrolled in school as girls from a community with no girls’ school nearby67, taking into account province, urban/rural location and vulnerability status of the household. This relationship was diluted because not all girls enrolled in school attended a government school and not all of the government girls’ schools identified were working effectively. Parents of children not attending school were asked the reasons why not. Some said it was because they could not

64 24121/43518 households within 1.5km of a government school were satisfied with government education compared with 926/2186 further from a school. OR 1.61, 95% CI 1.47-1.76, stratified by province, urban/rural and vulnerability 65 18794/25843 boys aged 5-9 years are enrolled in school compared with 12951/22876 girls. OR 2.09, 95% CI 2.01-2.17, stratified by province, urban/rural 66 5877/11747 children from very vulnerable households are enrolled compared with 25227/36062 children from less vulnerable households. OR 0.47, 95% CI 0.45-0.49, stratified by province, urban/rural, sex 67 10280/16896 girls in communities with a girls’ school nearby are enrolled compared with 2491/5712 without a girls’ school. OR 1.83, 95% CI 1.71-1.96, stratified by province, urban, vulnerability

CIET2003

Fig. 15 Net school enrolment of boys 5-9 years

CIET2003

Fig. 16 Net school enrolment of girls 5-9 years

43

afford it. The parents of non-school-going children from ‘very vulnerable’ households were more likely to say their reason was they could not afford the costs of sending the child to school, compared with parents from less vulnerable households68. The common reasons given for boys and girls not attending school are shown in Table 9. For both boys and girls, the commonest reason parents gave for the child not attending school was that they cannot afford it. Another common reason mentioned by parents was that the child was “too young” although all the children in the survey were at least five years old. There was a notable difference between boys and girls in the reason ‘no need to send to school’, mentioned for 9% of the boys and 20% of the girls. And for another 20% of the girls, the reason they were not going to school was the lack of a girls’ school or lack of female teachers in another school the girl could attend. For a non-school-going boy, the parents were only half as likely to give the reason ‘no need’ compared with a girl 69. The contrast between boys and girls was most marked in NWFP, where the reason ‘no need’ was given for 28% of non-school-going girls and only 8% of non-school-going boys.

Type of school attended Across the country, two thirds (67%) of the 5-9 year olds in school (n=31707) attended government schools. Nearly a third attended private schools (31%) and 3% attended non-formal education (NFE) or religious schools. The proportions of school-going children in government schools across the provinces are shown in Figure 17. The proportion attending a government school was highest in Balochistan at 87%, while in Punjab 42% of the enrolled children attended private schools. Among school-going children, the household vulnerability apparently made a difference to choice of school. A child from a ‘very vulnerable’ household was twice as likely to attend a government school compared with a child from a less vulnerable household70. The effect was most marked in

68 1970/5856 parents of non-enrolled children in very vulnerable households said it was because they could not afford it, compared with 2620/10791 parents in less vulnerable households. OR 1.98, 95% CI 1.76-2.05, after stratification by province, urban/rural 69 628/7020 non-school-going boys had the reason “no need” compared with 1939/9896 non-school-going girls. OR 0.42, 95% CI 0.38-0.46, after stratification by province. 70 5142/5866 children from very vulnerable households attended a government school compared with 17553/25199 from less vulnerable households. OR 2.16, 95% CI 1.98-2.36, stratified by province, urban

67

53

55

69

87

73

0 50

Pakistan

Isb

Punjab

NWFP

B'stan

Sindh

Fig. 17. % of enrolled children attending a government school

Table 9. Reasons for children aged 5-9 years not attending school

Reason Boys % (n)

Girls % (n)

Cannot afford 31(2202) 25(2464) Child too young 31(2176) 17(1666) No need/home study

9 (628) 20(1939)

No girls’ school / female teacher

- 20(2006)

No teachers 9 (641) 5 (521) Child handicapped

6 (428) 3 (263)

Studying religion 3 (240) 3 (285) Poor access 3 (231) 3 (277) Child working 3 (212) 1 (108) Poor teaching 2 (119) 2 (157) Poor facilities 1 (88) 2 (181)

44

Sindh. The relationship between vulnerability and going to a government school was also more marked in urban sites. Thus in urban sites of Sindh71 and NWFP72 school-going children of ‘very vulnerable’ households were much more likely to attend a government school, compared with children from less vulnerable households. The strong effect of vulnerability on school choice in these urban sites probably reflected the greater availability of private schools, so that parents who could afford to send their children to these private schools exercised this choice, while the poorest households were left with the cheaper option.

Parental satisfaction with schools Across the country, 73% (n=22890) of parents of children aged 5-9 years attending government schools said they were satisfied with the school, while 89% (n=7726)of parents whose children were attending a private school were satisfied, and 90% (n=766) of the small number attending a religious or non-formal were satisfied. The levels of parental satisfaction with different types of school were remarkably consistent across all the four provinces. Figure 18 shows the parental satisfaction with government schools. Parents of children attending a government school were only a third as likely to be satisfied with the school compared with parents whose children at attending private education of some sort73. Parents of boys in school were marginally more likely to be satisfied compared with parents of girls74. Parents of school-going children from ‘very vulnerable’ households were less likely to be satisfied with the children’s school compared with parents from less vulnerable households, but this effect was mainly because they were more likely to be in government schools. Government schools serving the sample sites were reviewed as part of the data collection (see Annex 9). If the school had an active PTA, the parents were more likely to be satisfied, compared with parents of a school without an

71 In urban sites of Sindh, 154/184 school-going children of very vulnerable households attend government schools compared with 976/2008 children of less vulnerable households. OR 5.43, 95% CI 3.56-8.31 72 In urban sites of NWFP, 89/114 school-going children of very vulnerable households attend government schools compared with 366/793 children of less vulnerable households. OR 4.15, 95% CI 2.54-6.83 73 17012/22887 parents of children in government schools were satisfied compared with 7632/8490 parents of children in private schools. OR 0.30, 95% CI 0.28-0.32, stratified by province 74 14692/18603 parents of boys in school were satisfied compared with 9979/12803 parents of girls. OR 1.07, 95% CI 1.01-1.13, stratified by province, urban/rural and vulnerability

73

77

70

72

77

81

0 50

Pakistan

Isb

Punjab

NWFP

B'stan

Sindh

Fig. 18. % parents satisfied with government schools

45

active PTA75. However, this difference was found only among the parents of boys, where parents in a school with an active PTA were 30% more likely to be satisfied compared with parents in a school without an active PTA76. Among girls the presence of an active PTA made no difference to parental satisfaction. This suggests that PTAs did not necessarily address the issues that concerned the parents of girls in school. Among girls, parents were 60% more likely to be satisfied if the school provided free text books for girls, compared with schools that did not provide textbooks77. Among boys, the provision of free text books for boys had much less effect on parental satisfaction78. Facilities provided by the schools had some effect on parental satisfaction. For example, parents were 17% more likely to be satisfied if the school had electricity than if it did not79.

Views of government school principals The head teachers of 751 government schools serving the survey communities were interviewed (see Annex 9). About half of them (51%, 380) thought the new local government plan would improve the educational service they could offer to the community (Table 10). The most common ways in which head teachers thought the new system would help to improve their services (among those who thought it would be better or unchanged) are shown in Table 11. ‘Better monitoring’ was the most common response. As an insight into practices in the system, heads were asked if they knew of any cases where teachers were asked to pay from their salary or to secure a post. Some 7% (55/744) knew of such cases and 6% (43/741) admitted it had happened to them.

75 5574/7349 parents of children in schools with an active PTA were satisfied compared with 5026/6749 in schools without an active PTA. OR 1.17, 95% CI 1.07-1.28, stratified by province, urban, vulnerability, sex 76 3539/4613 parents of boys in schools with an active PTA were satisfied compared with 2857/3956 in schools without an active PTA. OR 1.29, 95% CI 1.15-1.45, stratified by province, urban, vulnerability 77 1934/2320 parents of girls in schools giving free girl’s text books were satisfied compared with 2131/3013 in other schools. OR 1.61, 95% CI 1.32-1.95, stratified by province, urban, vulnerability 78 2245/2913 parents of boys in schools giving free boy’s text books were satisfied compared with 4010/5487 in other school. OR 1.15, 95% CI 1.02-1.31, stratified by province, urban, vulnerability 79 4490/5888 parents of children in schools with electricity were satisfied compared with 5917/7977 in schools without electricity. OR 1.17, 95% CI 1.07-1.27, stratified by province, urban/rural

For boys in government schools, parents were more likely to be satisfied if there was an active PTA. This association was not found for girls.

Table 10. Heads of government schools: the local government plan

Effect on services % (n) Make them better 51 (380) No effect 31 (233) Make them worse 4 (26) Don’t know 15 (112)

Table 11. Heads of government schools: effects of the local government plan

Response % (n) Better monitoring 31 (112) More facilities 25 (91) Problem solving 14 (49) Improved quality 11 (38) Provided LG works 5 (18) More enrolment 4 (15) More govt support 4 (15)

46

Water supply

Household satisfaction with government water supply Just 18% (n=46644) of households across the country reported they were satisfied with the government water supply, another 20% were not satisfied and nearly two thirds (62%) said they did not have access to a government water supply. The proportion of the households who reported satisfaction with the government water was notably low at 14% (n=16698) in Punjab, where 70% of the households reported they had no access to a government water supply. The proportions of households in each province satisfied with the government water supply and with access to a government water supply are shown in Figures 19 and 20. Male household respondents were somewhat less likely to report satisfaction with the government water supply compared with female respondents80. Urban households were more likely to be satisfied with the government water supply compared with rural households81. And urban households were much more likely to have access to a government water supply compared with rural households82. The difference was most marked in Sindh. Respondents from ‘very vulnerable’ households were less likely to be satisfied with the government water supply compared with those from less vulnerable households83. ‘Very vulnerable’ households were also less likely to report access to a government water supply compared with less vulnerable households84. The lower access to government water supply among ‘very vulnerable’ households was more marked in urban locations. Areas of towns with a high proportion of very poor households may not have been prioritized for provision of a water supply.

80 3611/20902 male respondents were satisfied with government water supply compared with 5964/25723 female respondents. OR 0.68, 95% CI 0.68-0.75, after stratification by province, urban/rural 814241/12386 urban households were satisfied with government water supply compared with 5337/34258 rural households. OR 3.24, 95% CI 3.08-3.41, after stratification by province. 82 8917/12386 urban households reported access to government water supply compared with 12853/34258 rural households. OR 6.70, 95% CI 6.36-7.06, after stratification by province 83 1180/7520 very vulnerable households were satisfied with government water supply compared with 8203/38287 less vulnerable households. OR 0.77, 95% CI 0.71-0.82, stratified by province, urban/rural 84 2920/7520 very vulnerable households reported access to government water supply compared with 18430/38287 less vulnerable households. OR 0.72, 95% CI 0.68-0.76, stratified by province, urban/rural

18

51

14

28

27

24

0 20 40 60

Pakistan

Isb

Punjab

NWFP

B'stan

Sindh

Fig. 19. % households satisfied with government water supply

38

85

30

60

68

47

0 50

Pakistan

Isb

Punjab

NWFP

B'stan

Sindh

Fig. 20. % households reporting access to government water supply

47

Type of water supply Across the country, the majority (78%, n=57071) of households reported a water supply (whether a government or private supply) inside their household or homestead. The proportion of households with their water supply inside the homestead ranged from 88% in Punjab (n=20824) to 58% (n=12189) in Balochistan (Figure 21). Urban dwellers were three times as likely to have their water supply inside the household compound compared with rural dwellers85. Respondents from ‘very vulnerable’ households were less likely to have their water supply inside the household compound compared with those from less vulnerable households86. This was true in both urban and rural settings. The pattern of water source nationally was that about a quarter (26%, n=57071) of households had a piped water source inside or outside the household, two thirds (66%) used some form of ground water (from pumps or wells), only 5% used surface water (springs, rivers, canals, ponds), and a few 3% got their water from tankers or vendors. Though Punjab had the highest proportion of households with water supply inside the household compound, only 20% (n=20824) of households used piped water; most households in Punjab (77%, n=20824) used some form of ground water. Balochistan had the lowest proportion (34%, n=12189) of households using a ground water source. The use of surface water sources (springs, rivers, canals, ponds) was higher in NWFP (16%) and Balochistan (14%) than elsewhere (Table 12).

Payments for water supply Across the country, more than four out of ten households (43%, n=56238) households paid for their water supply. Among those who paid, the mean amount per month was Rs362 overall. The proportion of households who paid for water was relatively high in Punjab (53%, n=20576) and in the federal capital (65%, n=459) (Table 13).

85 13131/154181 urban households have their water supply inside the compound compared with 26805/41590 rural household. OR 3.15, 95% CI 3.00-3.32, after stratification by province. 86 5060/8935 very vulnerable households have their water supply inside the compound compared with 34082/47085 less vulnerable households. OR 0.70, 95% CI 0.66-0.73, stratified by province, urban/rural

78

80

88

64

58

66

0 50 100

Pakistan

Isb

Punjab

NWFP

B'stan

Sindh

Fig. 21. % households with water supply within the homestead

Water sources nationally: • 26% piped water • 66% ground water • 5% surface water • 3% tankers/vendors

Table 12. Household water sources Type of water source: % HH Area Piped Grnd Surf Tank

Sindh 34 53 3 11 Bal’stan 46 34 14 6 NWFP 43 40 16 1 Punjab 20 77 3 0 Isl’bad 77 22 1 1 Pakistan 26 66 5 3

Grnd=ground water, surf=surface water, tank=tanker/vendor

Table 13. Payments for water supply Area % who

pay Mean mnthly cost (Rs.)*

Sindh 39 348 Balochistan 31 280 NWFP 31 366 Punjab 53 378 Islamabad 65 224 Pakistan 43 362

*Among those who paid anything for water

48

Urban households were over three times as likely to pay for their water supply compared with rural households87. Respondents from ‘very vulnerable’ households were less likely to pay for their water supply compared with those from less vulnerable households88. This was true for both urban and rural locations. Households with government water supply (essentially piped water) were more than twice as likely to pay for water compared with households using some other, non-government, water supply89. The association was particularly marked in NWFP and Balochistan.

87 8766/14937 urban dwellers paying for their water supply compared with 12448/41301 rural dwellers. OR 3.22, 95% CI 3.10-3.36, after stratification by province. 88 2074/8864 very vulnerable households paid for their water supply compared with 18704/46347 less vulnerable households. OR 0.59, 95% CI 0.56-0.63, stratified by province, urban/rural 89 10085/17635 households who used piped water paid for it compared with 11097/38491 households who used other water sources. OR 3.93, 95% CI 3.76-4.11, stratified by province, urban/rural

A rural water source. Photo: CIET

49

Perceptions and Experience of the Police / Administration Provision of a secure civil environment and maintenance of law and order are crucial administrative components of public services. In most of the areas of the country it is the responsibility of the police to undertake these functions. However in some areas of rural Balochistan and some parts of NWFP and Punjab, these functions are undertaken by levies working under local administration. In order to allow for these inter-provincial and regional differences in provision of services, the questions concerning the police and police functions were phrased and modified as necessary in different regions to ensure they covered the local situation. In the following sections the term ‘police’ is used to cover both the police and local levies.

Perceptions about the police / administration Household respondents were asked about who they would contact if they had a problem of personal safety or a problem of a threat to their property. They gave a variety of responses to each question (Tables 14 and 15). In both cases, the most common response was “Allah” or “only Allah could help me”. Police and community sources of help were the next most common choices for both male and female respondents. For both types of problem, about 5% of male and female respondents mentioned they would go to the nazim or a councillor for help. Across the country, 22% of household respondents (n=56778) said they would contact the police for a matter of personal safety, and 25% (n=56706) said they would contact the police about a threat to their property. The proportions of household respondents who would use the police for a problem of personal safety or a threat to property were lowest in Balochistan (14% and 15%) (Figure 22). Male respondents were more likely to say they would contact police for a problem of personal safety90 and or a threat to property91 compared with female respondents. The

90 6747/24512 men would contact the police for a problem of personal safety compared with 5091/32241 women, OR 1.95, 95% CI 1.86-2.03, stratified by province, urban/rural 91 6894/24490 men would contact the police for a threat to property compared with 6131/32191 women, OR 1.58, 95% CI 1.52-1.65, stratified by province, urban/rural

Table 14. Sources of help for a problem of personal safety: % (n)

Response Male Female Allah 39(9585) 46(14757) Police 28(6747) 16 (5091) Community source 9 (2259) 10 (3251) Family 4 (848) 12 (3776) Noone /not needed 8 (1999) 6 (1969) Nazim/councillor 5 (1274) 4 (1430) Government 4 (1008) 3 (874) Friends/neighbours 3 (602) 3 (933) Courts 0 (108) 0 (67)

Table 15. Sources of help for a threat to property: % (n)

Response Male Female Allah 35(8439) 39(12614) Police 28(6894) 19(6131) Community source 10(2381) 11(3642) Noone /self 9 (2071) 7 (2275) Family 3 (817) 11 (3382) Nazim/councillor 6 (1365) 5 (1566) Government 5 (1199) 4 (1261) Friends/neighbours 2 (572) 3 (811) Courts 3 (688) 1 (427)

22

54

20

26

14

26

0 20 40 60

Pakistan

Isb

Punjab

NWFP

B'stan

Sindh

Fig 22. % households who would use the police for a problem of personal safety

50

gender differential was present in all provinces, most marked in NWFP. Respondents from ‘very vulnerable’ households were less likely to say they would contact the police for a problem of personal safety92 or for a threat to property93. This was true in both urban and rural settings, and whether or not there was a police station in the community. Access to the police made a difference. Households in communities with a police station on site were more likely to say they would use the police for a problem of personal safety94 or a threat to property95 compared with those in communities without a police station. Similar associations were present considering communities with a police station within 5 km. However, this apparent intention to use the police for different issues did not necessarily translate into actual contacts with the police (see below). Across the country, nearly a third of households (30%, n=57183) say the police in their area make them feel safe. There is not much variation in this figure across the provinces: it ranges from 28% in Sindh and Punjab to 38% in NWFP. (Figure 23). Male household respondents were somewhat more likely to say the police made them feel safe compared with female respondents96. In most provinces there was little difference between men and women, but in Balochistan men were more than twice as likely to say the police (levies) made them feel safe compared with women97. ‘Very vulnerable’ households were less likely to say the police made them feel safe compared with less vulnerable households98. This was true in both urban and rural settings, and whether or not there was a police station in the community.

92 1251/8918 very vulnerable households would contact the police for a problem of personal safety compared with 10322/46807 less vulnerable households. OR 0.70, 95% CI 0.65-0.75, stratified by province, urban/rural and presence of police station in community 93 1393/8909 very vulnerable households would contact the police for a threat to property compared with 11353/46749 less vulnerable households. OR 0.70, 95% CI 0.66-0.75, stratified by province, urban/rural and presence of police station in community 94 1385/6303 households in sites with a police station would use the police for a personal safety problem compared with 7699/39638 in other sites. OR 1.31, 95% CI 1.23-1.41, stratified by province, urban, sex 95 1527/6291 households in sites with a police station would use the police for a threat to property compared with 8562/39609 in other sites. OR 1.36, 95% CI 1.27-1.46, stratified by province, urban, sex 96 8009/24569 male respondents say the police make them feel safe compared with 9672/32589 female respondents. OR 1.12, 95% CI 1.08-1.16, stratified by province, urban/rural 97 In Balochistan, 1479/3558 male respondents say the police make them feel safe compared with 2235/8646 female respondents. OR 2.04, 95% CI 1.88-2.22 98 2360/8962 very vulnerable households say the police make them feel safe compared with 14983/47164 less vulnerable households. OR 0.78, 95% CI 0.74-0.83, stratified by province, urban, sex, police station

30

36

28

36

29

28

0 20 40

Pakistan

Isb

Punjab

NWFP

B'stan

Sindh

Fig 23. % households who say the police make them feel safe

“The police are so threatening that one trembles just at the sight of them.” Female focus group, Badin

51

Households in communities with a police or levies station inside the community (see Annex 6) were about 30% more likely to say the police made them feel safe compared with households in communities without a station99, taking into account province and urban/rural location. There was a much weaker association between feeling safe through the police and a police or levies station within five kilometres of the community. In communities that had a police or levies station on site, ‘very vulnerable’ households were still less likely to report the police made them feel safe, compared with less vulnerable households100.

Experience of contacts with the police / administration Contacts with the police Only 12% of households (n=57006) across the country reported any contact with the police for any reason during the last five years. The proportion of households reporting such a contact was highest in Punjab (16%, n=20827) and lowest in Balochistan (5%, n=12139) (Figure 24). In the federal capital 14% (n=489) households reported a contact with the police. This pattern of contacts with the police was not explained by relative geographic access to the police. According to the community profiles of the sample sites, relatively more of the sample communities in Balochistan had a police or levies station within the community or within 5 km (see Annex 7 for information from the community profiles) than in other provinces. After stratifying by province, urban/rural location and sex of the respondent, there was no association between the presence of a police station in the community and contacts with the police in the last five years. There was a marked difference in reporting of household contacts with the police depending on the sex of the household respondent. A male household respondent was more than twice as likely to report a household contact with

99 2231/6330 households in sites with a police/levies station said the police make them feel safe compared with 11370/39911 households in sites without a station. OR 1.33, 95% CI 1.25-1.41, stratified by province, urban/rural location, household vulnerability 100 In sites with a police station, 317/1105 very vulnerable households said the police made them feel safe compared with 1863/5097 less vulnerable households. OR 0.70, 95% CI 0.60-0.81

12

14

16

8

5

7

0 5 10 15 20

Pakistan

Isb

Punjab

NWFP

B'stan

Sindh

Fig 24. % households with a police contact in the last five years

52

the police in the last five years compared with a female household respondent101. This probably reflected lack of awareness and involvement of women in such matters. The effect was more pronounced in Balochistan where a male respondent was almost four times more likely to report a household contact with police/administration during the last five years compared with a female respondent102. As expected, urban households were more likely to have had police contact in the last five years compared with rural households103. The urban/rural difference was particularly marked in Balochistan. ‘Very vulnerable’ households were less likely to have had a police contact in the last five years compared with less vulnerable households104. This was true in both urban and rural settings. For nearly all the reported contacts with the police in the last five years the involved household member was male (96%, n=4544). This very high proportion of males amongst police contacts was found in all areas of the country. The low proportion of females contacting the police limits the analysis by sex of the involved person. Also, in many cases, although the involved person was a woman, males from the household may have interacted with the police on her behalf. The main reported reasons for the contacts with the police are shown in Table 16. Domestic conflicts (including marital disputes, maintenance and child custody issue) were the main reported reason for the contact, followed by robbery and property rights. Among the small number of female contacts, there was a relatively higher proportion of domestic conflicts as the reason for the contact. A little under half the reported police contacts (46%, n=4570) were said to have been initiated by the police, as opposed to the household member(s). ‘Very vulnerable’ households were less likely to report their contact with the police was initiated by themselves (rather than by the police), compared with less vulnerable households105. This was true in both urban and rural locations. This suggests

101 3434/24510 male respondents reported a household contact with police during last five years compared with 2159/32471 female respondents, OR 2.17, 95% CI 2.04-2.29, stratified by province, urban/rural 102 In Balochistan 359/3545 households with male respondent reported a contact with police during last five years compared with 243/8580 households with female respondents, OR 3.87, 95% CI 3.25-4.60 103 1736/15422 urban households had police contact in the last 5 years compared with 3859/41582 rural households. OR 1.21, 95% CI 1.14-1.29, stratified by province 104 672/8927 very vulnerable households had police contact in the last 5 years compared with 4830/47029 less vulnerable households. OR 0.89, 95% CI 0.82-0.98, stratified by province, urban/rural 105 276/550 very vulnerable households initiated their police contact themselves compared with 2233/3953 less vulnerable households. OR 0.78, 95% CI 0.65-0.93, stratified by province, urban/rural

Table 16. Reasons for contacts with the police: % (n)

Reason Male Female Domestic conflict 29(1250) 42 (81) Robbery 19 (820) 15 (28) Property rights 16 (695) 14 (27) False charges 7 (303) 2 (4) Physical assault 7 (304) 8 (15) Traffic issue 6 (264) 4 (7) Murder 5 (199) 7 (13) Illegal arms/drugs 3 (109) 1 (1) Support for friend 2 (79) 2 (3) Missing person 1 (52) 4 (7) Political/tribal issue 2 (65) 0 Lost documents 1 (46) 0

Very vulnerable households were more likely to have their police contacts initiated by the police, compared with less vulnerable households.

53

either poor access to the police of the most disadvantaged citizens or a greater tendency of the police to contact these citizens, or both. The overall lower level of police contacts reported by ‘very vulnerable’ households (see above) suggests it was probably mainly a question of poor access to police services by these poor households. Registration of First Investigation Report (FIR) In about half the reported contacts with the police in the last five years (49%, n=5302) households reported that an FIR was registered for the case. The proportion of contacts with an FIR registered was highest in NWFP (60%, n=968). (Figure 25). Households in urban areas were less likely to report registration of FIR for their contact with the police compared with rural households106. This may reflect the different type of cases involved in urban and rural contacts with the police. An FIR was almost twice as likely to be registered if the contact with the police was initiated by the household member(s), compared with the contact being initiated by the police themselves107. This relationship persisted after stratification by urban/rural location, and was more marked in urban communities. The type of case was probably different between contacts initiated by households and by the police themselves. Perhaps cases where contact was initiated by the household were more likely to be of a kind that required registration of an FIR. ‘Very vulnerable’ households who had contact with the police were slightly more likely to report that an FIR was registered, taking into account who initiated the case, urban/rural location and province108. This suggests the type of cases they were involved in might be more likely to require an FIR. Satisfaction with police contacts Across the country, about a third of households (31%, n=5528) who reported a contact with police during the last five years were satisfied with the way police treated them on that contact. Reported satisfaction with police contacts

106 742/1636 urban households reported an FIR was registered for their contact with police compared with 1951/3656 rural households, OR 0.77 95% CI 0.68-0.87 stratified by province 107 1390/2457 cases with contact initiated by the household had an FIR registered compared with 804/1937 cases in initiated by the police. OR 1.83, 95% CI 1.62-2.07, stratified by province, urban/rural 108 1169/2207 very vulnerable households had an FIR registered for the police contact compared with 1487/3001 less vulnerable households. OR 1.10, 95% CI 0.97-1.25, stratified by province, urban location and who initiated the contact.

49

54

48

60

53

40

0 20 40 60

Pakistan

Isb

Punjab

NWFP

B'stan

Sindh

Fig 25. % police contacts with an FIR registered

31

34

29

40

40

31

0 20 40

Pakistan

Isb

Punjab

NWFP

B'stan

Sindh

Fig 26. % police contacts satisfied with the service

54

was higher in Balochistan and NWFP than in Sindh and Punjab (Figure 26). In Punjab the proportion of households with police contact was highest (see Figure 24), but the satisfaction with the contacts was the lowest (Figure 26). The reasons given for satisfaction or lack of satisfaction with the police contact are shown in Tables 17 and 18. The most common reasons given by those who were satisfied were that their problem was solved, or that the police had a good attitude or were helpful. Among those who were dissatisfied, the most common complaints were about payments and bad attitude. Households who themselves contacted the police were nearly twice as likely to say they were satisfied with the service compared with households who were contacted by the police109. The effect remained after taking into account other variables. In male and female community focus group discussions, participants talked about what improvements they would like to see in the police. Many felt that better recruitment policies for the police, ensuring honest people were recruited, would help. Some groups voiced the opinion that if police salaries and incentives were better this would improve the service, citing the motorway police as an example of how this policy could work.

Citizen monitoring of the police In focus groups participants discussed about what a CCB could do in order to monitor the police. Many of the groups expressed concerns about citizens monitoring the police. The focus groups of women particularly voiced this opinion. However, after some discussion, many groups made suggestions about how they as citizens, perhaps as members of a CCB, could monitor the police. The most common suggestion was for direct observation to check attendance and behaviour. Another common idea was to form a special group or appoint people with special responsibility to monitor the police. Some focus groups mentioned asking people who had been in contact with the police about their experience.

109 967/2532 households who themselves contacted the police were satisfied compared with 495/2010 households who were contacted by the police. OR 1.60, 95% CI 1.38-1.85, stratified by province, urban/rural, payment, if FIR registered

Table 17. Reasons for satisfaction with police contact Reason % (n) Problem solved 33 (595) Good attitude 23 (417) Police helpful 21 (372) Got justice 13 (238) Had connections 5 (86) Paid so got service 4 (74) Got good advice 1 (22)

Table 18. Reasons for dissatisfaction with police contact

Reason % (n) Payments 26 (944) Bad attitude 21 (768) Did not help 14 (522) Problem not solved 11 (405) Paid to get service 7 (254) No justice 7 (250) Made things worse 6 (208) Favoured other party 4 (146) Had to use connections 2 (81)

55

Perception and Experience of Courts

Opinions about the courts Nationally 46% of household respondents (n=57130) thought the courts were there to help them. The proportion was lowest in Balochistan where only 33% (n=12205) of the householders held this view (Figure 27). The reasons given by respondents for their views about whether or not the courts were there to help them are shown in Tables 19 and 20. The most common reason for thinking the courts were there to help was that it was “their duty”. The most common reason given for thinking the courts were not there to help was that ‘you need money’. Male household respondents were more likely to say the courts were there to help, compared with female respondents, but this was mainly because they were more likely to give a definite opinion. Over 40% of female household respondents could not say if the courts were there to help or not. Urban households were also apparently more likely to think the courts were there to help, but again this was mainly because fewer rural household respondents were able to express an opinion about this issue (36% could no give an opinion). The same was true for ‘very vulnerable’ compared with less vulnerable households: over 40% of the ‘very vulnerable’ households could not give an opinion about the courts.

Experience of contacts with the courts Nationally, only 8% (n=56887) of households reported a contact with the courts during the last five years. Contacts were especially infrequent in Sindh (3%) and Balochistan (3%) (Figure 28). Male household respondents were more than twice as likely to report a household contact with the courts during the last five years compared with female respondents110. The difference was especially marked in Balochistan, where male respondents were more than four times more likely to report a court contact in the household compared with female respondents. This differential reporting rate

110 2307/24426 male respondents reported a household court contact in the last 5 years compared with 1381/32436 female respondents. OR 2.13, 95% CI 1.99-2.29, stratified by province, urban/rural

46

48

48

49

33

44

0 20 40 60

Pakistan

Isb

Punjab

NWFP

B'stan

Sindh

Fig 27. % households who think the courts are there to help them

Table 19. Reasons why the courts are there to help: % (n)

Reason Male Female It’s their duty 66(8125) 61(7641) Helpful / trustworthy 24(2950) 29(3575) Never contacted 3 (319) 3 (316) If you have money 3 (313) 2 (300) If have connections 2 (231) 2 (231)

Table 20. Reasons why the courts are not there to help: % (n) Reason Male Female You need money 50(3092) 41(2400) No justice 28(1698) 30(1804) Not helpful 7 (422) 11 (675) Never contacted 6 (347) 11 (674) Takes too long 8 (462) 4 (261)

8

8

9

8

3

3

0 5 10

Pakistan

Isb

Punjab

NWFP

B'stan

Sindh

Fig 28. % households who contacted the courts in the last five years

56

probably reflected the lack of involvement of women in these important issues. There was no difference between urban and rural households in their contact with the courts in the last five years. ‘Very vulnerable’ households, however, were less likely to have had a court contact compared with less vulnerable households111. Geographic access did make some difference. Households in communities within 10 km of a court (see Annex 7) were about 15% more likely to have had contact with the courts in the last five years compared with households in communities with the nearest court more than 10 km away112. The great majority of reported court contacts in the last five years were for male household members (92% n=2930). This was true across all the provinces. The main reported reasons for the court contacts are shown in Table 21. Among women there were relatively fewer cases for property rights and more for domestic disputes, compared with men. Satisfaction with the courts Nearly half of the households (48%, n=3633) who reported contact with the courts during last five years said they were satisfied with the way they were treated. There was very little inter-provincial variation. The proportion of court users satisfied was apparently lower (35%, n=49) among the small number of court contacts reported from the federal capital. There was no difference in satisfaction with the courts by gender, although this was difficult to examine because of the very small number of women with court contacts. People from ‘very vulnerable’ households in contact with the courts were somewhat less likely to be satisfied with the experience compared with people from less vulnerable households113. The reasons given for satisfaction with court contacts are shown in Table 22. The main reason was the court user felt

111 400/8923 very vulnerable households contacted the courts in the last 5 years compared with 3229/46916 less vulnerable households. OR 0.80, 95% CI 0.72-0.89, stratified by province, urban/rural 112 1357/19618 households in sites within 10km of a court used the courts in the last 5 years compared with 1556/26410 in communities further from a court. OR 1.15, 95% CI 1.05-1.25, stratified by province, urban/rural and sex of respondent 113 177/394 from very vulnerable households were satisfied with the court contact compared with 1560/3170 from less vulnerable households. OR 0.82, 95% CI 0.66-1.01, stratified by province, urban/rural

Table 22. Reasons for satisfaction with court contact Reason % (n) Got justice 59 (1018) Problem solved 18 (310) Good behaviour 11 (195) Court helpful 10 (168)

Table 21. Reasons for court contacts: % (n) Reason Male Female Property rights 47 (1241) 28 (59) Domestic dispute 19 (499) 51 (106) Murder 6 (160) 5 (11) Robbery 5 (143) 0 Physical assault 5 (120) 2 (5) False custody 3 (84) 1 (3) Traffic issue 2 (63) 1 (2)

57

they “got justice”. The main complaints among those who were dissatisfied (Table 23) were about payments or the length of the process.

Alternative mechanisms Few households had heard of the reconciliation committees at union council level intended for settling small disputes without recourse to the courts. Nationally, only 7% (n=56961) of households were aware of reconciliation committees. The proportion who had heard of reconciliation committees ranged from 3% in Balochistan, through 4% in NWFP and 6% in Sindh and 9% in Punjab. Men were twice as likely to have heard of reconciliation committees compared with women114. Urban households were nearly twice as likely to have heard of reconciliation committees compared with rural households115. ‘Very vulnerable’ households were less than half as likely to have heard of reconciliation committees compared with less vulnerable households116. This was true in both urban and rural settings.

114 1954/24457 male respondents had heard of reconciliation committees compared with 1205/32479 female respondents. OR 2.27, 95% CI 2.10-2.45, stratified by province, urban/rural 115 1306/15414 urban households had heard of reconciliation committees compared with 1853/41547 rural households. OR 1.84, 95% CI 1.70-1.98, after stratification by province 116 196/8938 very vulnerable households had heard of reconciliation councils compared with 2876/46981 less vulnerable households. OR 0.46, 95% CI 0.39-0.53, stratified by province, urban/rural

Table 23. Reasons for dissatisfaction with court contact Reason % (n) Payments 41 (753) Took too long 35 (632) Made problem worse 17 (303) Bad behaviour 5 (83)

58

Local government

Voting in local council elections 2001 Nearly all (94%, n=46700) the household respondents in the 2002 main survey had heard about the 2001 local council elections. Male respondents were more likely to have heard about the local council elections compared with female respondents117. The 2002 survey included questions about the number of men and women in the household who voted. Nationally, most (89%, n=42052) households reported that at least one male member voted in the 2001 local council elections, and 72% (n=42319) reported that at least one female member voted. There was considerable variation across the country (Figure 29). The lowest proportion of households with at least one man voting in the 2001 elections was in Balochistan (62%, n=8336). The lowest proportion of households with at least one woman voting was in NWFP (43%, n=10134). The gender gap between male and female voting was notably high in NWFP, with the highest proportion of households with at least one man voting (91%) and the lowest proportion with at least one woman voting (43%). In Balochistan the proportion was low for both male and female voting (Figures 30 and 31). A similar picture emerged when considering the mean percentage of male and female household members who voted in the 2001 local council elections. Across the country, the mean proportion of male household members who voted was 77% (n=42026) and of female household members who voted 62% (n=42294). The mean percentage was low for both men (55%) and women (45%) in Balochistan and there was a marked gender gap (78% men and 35% women voted) in NWFP. There was little relationship between household vulnerability and male voting. However, ‘very vulnerable’ households were less likely to have at least one woman voting118. This association remained after stratification by

117 20533/20952 male respondents had heard about the 2001 local council elections compared with 23336/25729 female respondents. OR 4.45, 95% CI 4.00-4.95, stratified by province, urban/rural 118 4173/6831 very vulnerable households had at least one woman voting compared with 22418/34884 less vulnerable households. OR 0.91, 95% CI 0.86-0.96, stratified by province, urban/rural

89

90

91

62

85

72

80

43

53

71

0 20 40 60 80 100

Pakistan

Punjab

NWFP

B'stan

Sindh

Women

Men

Fig 29. % households with at least one man or woman voting in 2001 elections

CIET2003

Fig 30. % households with at least one man voting in 2001 elections

CIET2003

Fig 31. % households with at least one woman voting in 2001 elections

59

urban/rural location and province but was found only in NWFP and Balochistan. There is thus some evidence that in some places the most disadvantaged groups were less than fully represented in the local democratic process of the union council elections 2001. The exclusion of women seemed to be more marked in the more disadvantaged households. This was seen mainly in NWFP and Balochistan.

Views about new union councils Nationally, expectations of the new union councils were mixed. Some 33% (n=57052) of household respondents said they thought the new union councils would be better than the previous system, and 36% thought they would not be better. Some 31% of household respondents could not say if the new union councils would be better or not. Male household respondents were more likely to offer a definite expectation (better or not) about the new union councils compared with female household respondents119. In three out of the four provinces, more households expected the new union councils to be better than not (Figure 32). Only in Punjab did more households say they thought the new system would not be better. Among households who expected the new union council to be better than the previous system, by far the commonest reason given was that it was already doing a good job and helping them (Table 24). Households who thought the new union council would not be better most commonly said that it was not apparently doing any good at the moment (Table 25). They were not prepared to give the new system the benefit of the doubt. Thus citizens were cautious about the new system, perhaps skeptical from previous reforms, but many were willing to give it a chance before passing judgment.

119 19127/24522 male respondents had a specific opinion about the new union councils compared with 19242/32505 female respondents. OR 2.35, 95% CI 2.26-2.44, after stratification by province

0% 20% 40% 60% 80% 100%

Pakistan

Punjab

NWFP

B'stan

SindhBetter Not better DK

Fig 32. Household expectations about the new union councils

Table 24. Reasons for expecting the new union council to be better: % (n)

Reason Male Female Doing good job 72(8165) 76(6434) Own people there 10(1181) 6(516) Just, honest 9 (963) 6 (522) Hope will be better 5 (516) 8 (659) Too early to say 3 (329) 3 (217) We will monitor 2 (206) 2 (132)

Table 25. Reasons for expecting the new union council not to be better: % (n)

Reason Male Female Not doing any good 84(6367) 87(9063) Corrupt, get money 10 (792) 8 (809) Too early to say 3 (247) 3 (313) Structure wrong 2 (111) 1 (111) Don’t trust govt 1 (68) 1 (151)

60

Contacts with union council members There was a difference between male and female household respondents when reporting whether someone in the household had contacted a member of the union council. Male respondents were more than twice as likely to mention a household contact with a union council member compared with female respondents120. Nationally, 21% (n=20745) of male respondents and 10% (n=25443) of female respondents reported a household contact with a union council member. It is likely that female respondents were not aware of all contacts. The actual level of household contacts with union councillors was probably close to that reported by male household respondents. Across all four provinces, the proportion of households with contacts with union councillors was notably lower when reported by female compared with male household respondents (Figure 33). Respondents who reported that a member of the household had contacted a member of the new union council in their area were asked the purpose of the contact. Only contacts made with a councillor in his or her role on the council were included. Some households mentioned social contacts with the new councillors, which is to be expected given that the councillors were residents of the same area where they acted as councillors. The main reasons for contacting a councillor are shown in Table 26. There was quite a wide spread of issues, depending on the particular concerns of different areas. The reasons mentioned by male and female respondents were largely similar, although women seemed to mention financial support rather more often than men did. Urban dwellers were more likely to report a household contact with union council member compared with rural dwellers121. It was encouraging to note that ‘very vulnerable’ households were only marginally less likely to report a contact with a union council member compared with less

120 4522/20875 male respondents reported household contact with a union councillor compared with 2455/25597 female respondents. OR 2.62, 95% CI 2.48-2.76, stratified by province, urban/rural 121 2084/12274 urban households reported contact with a union council member compared with 4896/34217 rural households. OR 1.25, 95% CI 1.18-1.32, after stratification by province.

21

21

23

23

22

10

11

8

9

9

0 10 20 30

Pakistan

Punjab

NWFP

B'stan

Sindh

Female

Male

Fig. 33. % households in contact with a union councillor, by sex of respondent

Table 26. Reasons for contacting a union council member: % (n)

Reason Male Female Financial support 12 (519) 27 (638) Water problem 17 (764) 15 (362) ID card/certificates 17 (736) 14 (327) Roads/streets 12 (545) 6 (146) Electricity 10 (452) 8 (184) Sewage/drains 8 (357) 7 (154) Community dispute 6 (279) 5 (118) Property / land 3 (136) 4 (97) Police issues 3 (151) 3 (72) Garbage 4 (181) 2 (46)

61

vulnerable households122. However, in rural areas, the difference between ‘very vulnerable’ and less vulnerable households in their contacts with councillors was more marked. A challenge for the new local government system is to ensure that the needs of the most disadvantaged members of society are fully served. Households that reported at least one man voting in the local council elections 2001 were more likely to report contact with a union councillor compared with households where no man voted123. Similarly, households where at least one woman voted were more likely to have contact with a union councillor compared with households where no woman voted124. This suggests that those households that engaged with the local democratic process by voting continued to engage with the local government structures in order to help solve their problems. Satisfaction with union councillor contacts Just over half (54%, n=6910) of households who reported a contact with a union council member were satisfied with the service they received. Satisfaction was higher in Sindh and Balochistan than in NWFP and Punjab (Figure 34). Urban households were more likely to report satisfaction in their contacts with union council members compared with rural households125. Male household respondents were more likely to report satisfaction with union councillor contacts compared with female respondents126. Respondents from ‘very vulnerable’ households were less likely to report satisfaction with union councillor contacts compared with respondents from less vulnerable households127, although the difference was only apparent in urban settings. This suggested that the most disadvantaged members of society were not necessarily having their needs and concerns addressed by the new local government structures.

122 1039/7551 very vulnerable households reported contact with a councillor compared with 5809/38112 less vulnerable households. OR 0.95, 95% CI 0.89-1.02 123 5893/35388 households where a man voted contacted a union councillor compared with 833/6705 households with no man voting. OR 1.39, 95% CI 1.28-1.51, stratified by province, urban/rural 124 4647/27033 households where a woman voted contacted a union councillor compared with 2054/15324 households with no woman voting. OR 1.36, 95% CI 1.28-1.44, stratified by province, urban/rural 125 1302/2065 urban households were satisfied with contact with union councillor compared with 2612/4845 rural households. OR 1.35, 95% CI 1.21-1.50, after stratification by province. 126 2643/4496 male respondents reported satisfaction with union councillor contacts compared with 1270/2411 female respondents. OR 1.40, 95% CI 1.26-1.56, stratified by province, urban/rural 127 541/1028 very vulnerable households were satisfied with their union councillor contact compared with 3297/5751 less vulnerable households. OR 0.77, 95% CI 0.67-0.88, stratified by province, urban/rural

54

49

55

70

69

0 20 40 60 80

Pakistan

Punjab

NWFP

B'stan

Sindh

Fig. 34. % households satisfied with union councillor contacts

62

Views of union nazims and union council members In each of the communities in the 2002 survey, the field teams interviewed members of the union council. Where possible the union nazim or naib nazim was interviewed. Also they attempted to interview at least one woman councillor as well as male councillors in some cases. In total, some 757 nazims, naib nazims and councillors were interviewed. All the nazims and naib nazims were men. Up to three people, and occasionally four people, were interviewed from a union council, depending on who was available at the time of the visit. Tabulations of their responses are given in Annex 10. Information needs The councillors mentioned the problems they considered to be priorities in their areas and two thirds of them (69%, 518/751) thought they already had enough information to allow them to deal with these priority problems. Nazims and naib nazims were nearly six times more likely to think they had all the information they needed compared with other council members128. Among ordinary union council members, men were more than twice as likely to think they had all the information they needed compared with women129. Responses about what information they already had and what further information they might need demonstrated some confusion among nazims and councillors about the sort of information that would be needed to guide interventions to solve problems. When asked how they would go about getting further information about a problem, the most common response was “from our seniors” Few councillors or nazims mentioned seeking information from the population. Involving citizens and CCBs Two thirds (68%, 514/753) of the nazims and councillors said they had a method of seeking the views of citizens in their area. Nazims and naib nazims were more likely to say they had a method of seeking citizen views compared with

128 222/249 nazims and naib nazims thought they had all the information they needed compared with 296/502 other councillors. OR 5.75, 95% CI 3.68-8.99, stratified by province, urban/rural 129 168/243 men thought they had all the information they needed compared with 128/259 women. OR 2.46, 95% CI 1.67-3.61, stratified by province, urban/rural

Union councils information base 156 nazims 94 naib nazims 243 male councillors 264 female councillors

63

ordinary councillors.130 Among ordinary councillors, men were more likely to say they had a way of seeking citizen’s views compared with women131. The most common methods mentioned were Mohalla committees (89%), Mohalla meetings (40%) and house visits (38%). At the time of the interviews with the union councillors, in the second half of 2002, about half of them (49%, 371/752) had heard of CCBs. Only 18% of the women councillors had heard of CCBs. Some 12% (87/751) said they had a CCB in their union council. This does not mean there were CCBs in 12% of the union councils in the survey, as some of the councillors were from the same union councils. At the time of the survey in 2002, 8% of the sample communities were said to have a CCB in place. The councillors were generally sanguine about setting up and running CCBs. Only 12% (10/83) reported any difficulties setting up existing CCBs and only 27% (161/588) foresaw any difficulties setting up CCBs in the future. Asked what they thought were the qualities or skills needed for CCB members, they most commonly mentioned honesty (79%) and education (59%). Training and training needs Almost all (95%, 719/757) the nazims, naib nazims and councillors claimed to have received training for their role as a nazim or council member. The sources of the training, among the 600 who reported this, are shown in Table 27. Most of the training was through the district government. Some of it was topic specific training rather than about the role of a councillor. Since 2002 there has been a lot of training for councillors various sources. In particular, a number of initiatives have provided training for women councillors, at both district and union council levels.. Only about half (51%, 343/669) the nazims and councillors considered they needed further training. This was not different between nazims and ordinary councillors or between male and female councillors. The further training requested was mostly covering technical and management issues.

130 216/249 nazims and naib nazims said they had a way to seek views of citizens compared with 298/504 ordinary councillors. OR 4.43, 95% CI 2.93-6.68, stratified by province, urban/rural 131 159/242 male councillors said they had a way to seek citizens’ views compared with 139/263 women councillors. OR 1.87, 95% CI 1.27-2.76, stratified by province, urban/rural

Table 27. Training reported by union nazims and councillors

Source % (n) District government 84 (505) Provincial government 10 (60) Federal government 5 (30) University 0.7 (4) International agency 0.2 (1)

In 2002, only 18% of women union councillors had heard of CCBs.

64

Citizen Community Boards (CCBs)

Participation in pre-existing voluntary groups Reported household membership of voluntary groups was low: just 2% of households with a male participating in a voluntary group and 0.6% with a female participating in a voluntary group. Female membership of voluntary groups was virtually confined to those households with a male participating in a voluntary group: only 106 households across the whole country reported a female member of a voluntary group but no male member in such a group. There was no significant difference between urban and rural settings in male or female membership of voluntary groups. But ‘very vulnerable’ households were less likely to have a male132 or female133 member of a voluntary group compared with less vulnerable households. Some 5% of the sample communities nationally were said to have women’s groups working in them. Households in these communities were more likely to have a woman member of a voluntary group compared with households in other communities134. However, even in the communities with women’s groups in place, only 1% of households reported any female members belonging to voluntary groups. A quarter of the sample communities were noted to have NGOs or CBOs working there (see Annex 7). In communities where there were CBOs or NGOs working, households were more than twice as likely to have members involved in a voluntary group compared with households in other communities135. Even in communities with active CBOs or NGOs, only 4% of households were involved in voluntary groups. This low level of engagement in voluntary community groups, especially among the most disadvantaged in the

132 94/7559 very vulnerable households have a male voluntary group member compared with 915/38350 less vulnerable households. OR 0.48, 95% CI 0.38-0.59, after stratification by province 133 24/7560 very vulnerable households have a female voluntary group member compared with 264/38361 less vulnerable households. OR 0.41, 95% CI 0.27-0.63, after stratification by province 134 31/2927 households in sites with women’s groups active had any woman involved in a voluntary group compared with 263/43346 in sites without women’s groups. OR 1.57, 95% CI 1.07-2.30, stratified by province and urban/rural 135 544/13868 households in communities with NGOs/CBOs had a member in a voluntary group compared with 566/32372 households in other communities. OR 2.34, 95% CI 2.06-2.65, stratified by province, urban

65

population, is one of the problems that the devolution of local government is intended to address. To be successful, the new Citizen Community Boards (CCBs) will need to do better than existing community groups at drawing in people from all sections of the community.

Willingness to participate in CCBs Very few household respondents had heard of CCBs at the time of the survey (mainly during 2002, but with the ten pilot districts covered in 2001). Nationally, only 2.9% of male respondents and 1.3% of female respondents had heard of CCBs. Figure 35 shows the variation in awareness about CCBs across provinces. In 8% of the sample communities nationally in 2002 there was reported to be already at least one CCB in existence. In these communities there was more awareness of CCBs. In communities with a CCB households were twice as likely to have heard of CCBs compared with communities without CCBs136. However, even where there was a CCB in existence, only about 4% of households on average had heard of CCBs. Interviewers gave the household respondents a brief description about CCBs. Following this explanation, there was definite interest from households in joining a CCB. Nationally, half (50%, n=24519) of male respondents were interested to join a CCB, but less than a third (29%, n=32416) of female respondents were willing (Figure 36). Male household respondents are more than twice as likely to be willing to serve on a CCB compared with female respondents137. The willingness to join a CCB among male and female respondents varied across the provinces, as shown in Figure 36 and Figures 37 and 38. The greatest interest among men was found in NWFP, but here the gender gap was the most marked. Respondents from urban households were somewhat less likely to be willing to serve on a CCB compared with respondents from rural households138.

136 113/3044 households in communities with a CCB had heard of CCBs compared with 782/43103 in communities without a CCB. OR 2.12, 95% CI 1.72-2.61, stratified by province and urban/rural 137 12983/24519 male respondents were willing to join a CCB compared with 9163/32416 female respondents. OR 2.72, 95% CI 2.63-2.83, after stratification by province 138 5413/15445 respondents from urban households would join a CCB compared with 16748/41515 from rural households. OR 0.86, 95% CI 0.82-0.89, after stratification by province

3

3

4

3

3

1

1

1

1

2

0 1 2 3 4 5

Pakistan

Punjab

NWFP

B'stan

Sindh

Female

Male

Fig. 35. % household respondents who have heard of CCBs

50

49

66

51

45

29

30

32

25

30

0 20 40 60 80

Pakistan

Punjab

NWFP

B'stan

Sindh

Female

Male

CIET2003

Fig. 36. % household respondents who would be willing to join a CCB

Fig. 37. % male respondents who would be willing to join a CCB

66

It is encouraging to find that respondents from ‘very vulnerable’ households were as willing as those from less vulnerable households to join a CCB. This was true across all the provinces and in urban and rural settings. Indeed, in urban settings, respondents from ‘very vulnerable’ households were more likely to be willing to join a CCB compared with respondents from less vulnerable households139. This finding suggests that CCBs indeed have the potential to engage even the most disadvantaged section of male society. However, even at the level of willingness to participate there was clearly a problem about involving women which will have to be addressed if the CCBs are to be really representative.

Community views about CCBs In the male and female focus groups held in each of the survey communities, the facilitator shared the information about the proportion of household respondents in the district had said they would be willing to join a CCB. He or she then asked the participants if they thought people from that community would be willing to participate as CCB members. Overall, some three quarters of the focus groups responded positively on behalf of the people in the community. Male groups more often responded positively than female groups. It is probable that in each case the groups were responding mainly on behalf of people of their own sex in the community. There was no difference in the proportion of male household respondents willing to join a CCB between communities where the male focus group said the community would be willing to join CCBs and those where the focus group thought people would not be willing. But for women the community “ethos” did make a difference. In communities where the female focus group thought people would be willing to join CCBs, women household respondents were nearly 50% more likely to say they would be willing to join a CCB, compared with communities where the female focus group was less optimistic140.

139 In urban sites, 469/1200 respondents from very vulnerable households would join a CCB compared with 4820/9011 from less vulnerable households. OR 1.20, 95% CI 1.06-1.36 140 5044/15441 female household respondents were willing to join a CCB in communities where the female focus group was positive about joining CCBs, compared with 2078/8672 in communities where the focus group was not positive. OR 1.45, 95% CI 1.36-1.54, stratified by province, urban/rural

Fig. 38. % female respondents who would be willing to join a CCB

CIET2003

“CCBs are like sight given to a blind person.” Male focus group, Kacchi

67

How to promote CCBs The focus groups mentioned a number of ways they thought would be useful to make people aware about CCBs. The common suggestions included: • Hold community meetings. This was the most common

suggestion throughout the country and from both male and female groups.

• Use electronic media (TV and radio). This suggestion came up in many of the focus groups, males more than females. It was especially mentioned by groups in Sindh (including those in Karachi).

• Through local government representatives. This was quite a common suggestion everywhere, meaning that people would like to hear about the CCBs from their elected representatives. This emphasizes the importance of involving elected representatives and other local government officials in publicity campaigns about CCBs, so that people appreciate how CCBs are related to the local government structures.

• Through the mosque or Pesh Imam. This suggestion was quite common in groups in all parts of the country, and especially so in Punjab. It was mentioned rather less often by groups in Balochistan.

• Through the newspapers. This was mentioned quite regularly by groups across the country.

• Through respected community members. This was frequently mentioned in all areas, perhaps especially so in Balochistan.

• Special campaigns. These were mentioned by some groups, but not very commonly. However, some of the other methods mentioned would include some sort of special campaign.

• Posters/banners/pamphlets. These were mentioned but not very frequently. This is interesting as such methods often form a major part of publicity campaigns.

The general picture that emerged was that people would generally prefer personal contacts of some sort – meetings or presentations from elected representatives or other trusted people – rather than more general “campaigns”. How to set up and run CCBs About half the focus groups said they did not expect there to be any problems setting up and running CCBs. The commonly mentioned predicted problems included: difficulty in finding suitable people to be members, political interference or resistance, lack of community interest, and lack of funds.

“It’s only the first step that is difficult to take and so it is with the establishment of CCBs.” Male focus group, Sibi

“CCBs are a check on influential people who usually get benefits. Therefore they are not happy with the establishment of CCBS.” Female focus group, Badin

68

In order for CCBs to be successful, by far the most common requirement mentioned (by nearly all groups) was the selection of suitable people. Suitable people were described as honest, educated and with skills necessary to monitor and develop projects. Groups also mentioned the importance of the CCBs comprising truly local people. Another key issue commonly mentioned was that the CCBs should have the necessary official recognition and authority. The need for adequate funding was mentioned also, but not as often as the other issues. The most common resource requirement for CCBs was said to be adequate funds. Groups also commonly mentioned resources like office space, furniture and transport. Some focus groups thought CCBs would need support staff . Nearly all the focus groups thought the government would be the source of the resources for the CCBs. However, groups quite commonly mentioned other options, including donations from well-off community members and community contributions.

“CCB members should work together like a sports team play a match.” Female focus group, Lahore

“One needs two hands to clap” Male focus group, Peshawar (referring to government and community partnership)

69

Conclusions The findings described in this report are intended to stimulate discussion. There is more that could be done in analysis to examine the inter-play of the factors related to outcomes such as citizen satisfaction, school enrolment, payments to the police, or contact with union councillors. Discussions of the findings and analysis to date with concerned parties will help to guide areas for further investigation, both within the baseline data set itself and in the repeated cycles of data collection that are planned. This is work in progress in the sense that now the findings need to be put to use by sharing them at different levels and using them to stimulate and guide the development of action plans by coalitions of planners, service providers and citizens. This process is presently ongoing in the focus district of Lasbela but needs to happen elsewhere as well. The litany of disadvantage of ‘very vulnerable’ households in relation to public services was striking.

Very vulnerable households: • Had less access to basic amenities like roads, sewage, electricity • Had less access to health services • Made more use of government rather than private health services • Did not pay less for government health service contacts • Were less satisfied with government health service contacts • Were less likely to know how to complain about health services • Were less satisfied with available government schools • Were less likely to enroll their children in school • Were more likely to unable to afford to enroll children in school • Were more likely to send their children to government schools • Were less likely to have/be satisfied with a government water supply• Were less likely to pay for water • Were less likely to use the police for problems of personal safety • Were less likely to use the police for threats to property • Were less likely to say the police make them feel safe • Were less likely to have contacted the police in the last five years • Were more likely to have contacts initiated by the police • Were more likely to have an FIR registered in a police contact • Were less likely to have contacted the courts in the last five years • Were less likely to be satisfied with court contacts • Were less likely to have heard of reconciliation courts • Were less likely to vote (female members) • Were less likely to have contacted a union councillor (rural areas) • Were less likely to be satisfied with councillor contacts (urban areas)• Were less likely to belong to voluntary groups • Were NOT less likely to be willing to join a CCB

70

This information about the situation of ‘very vulnerable’ households sets out the challenge: to ensure that the new processes at local level are fully accessible to the most disadvantaged and that public services become more accessible to them and more responsive to their needs. The benchmark laid down here can be used to assess progress over time and in different parts of the country. The question is not only whether services and satisfaction improve on average, but whether the gap between the most disadvantaged and the rest of society closes as intended. The information on perceptions, use and experience of public services disaggregated by gender provides important baseline information for gender-responsive budgeting. The findings also set a benchmark for assessing progress in gender equality under devolution.

Within households women were often excluded from decisions and discussions about matters that nevertheless concern them. They were sometimes unaware that members of the household had contacted the police or a union councillor. Especially in some parts of the country, women were discouraged or even prevented from voting in the local council elections, despite the provision for a third of seats to be filled by women. Outside the household, women were limited in what they could do. Their participation in voluntary groups was limited, again especially in some areas. It is of concern that women were only half as likely as men to say they would be willing to join a CCB, with this gender gap being much greater in

Women household respondents: • Were much less likely to have basic education • Were less likely to express dissatisfaction with available amenities,

such as sewage, roads, electricity • Were less likely to know how to complain about health services • Were less likely to use the police for personal safety • Were less likely to use the police for a threat to property • Were less likely to say the police make them feel safe • Were less likely to report household contacts with the police • Were much less likely to have a police contact in the last 5 years • Were less likely to report household contacts with the courts • Were much less likely to have a court contact in the last 5 years • Were less likely to have heard of reconciliation committees • Were less likely to have voted in the local council elections • Were less likely to report household contact with union councillors • Were less likely to report satisfaction with union council contacts • Were less likely to belong to voluntary groups • Were only half as likely to be willing to join a CCB Girls: • Were half as likely as boys to be enrolled in school

71

some areas. Discussion in the focus groups of women confirmed that they would have to take permission to participate in CCBs and that this permission might not be forthcoming. This lack of full participation of women in community life is perpetuated in part by the lack of education of women, with the gender gap in primary school enrolment being still very apparent in 2002. Again this information highlights a challenge for devolution and the new institutions for increasing citizen engagement. If CCBs are to be truly representative they need to include women and the indications so far are that this will not happen without additional efforts. This can range from things as simple as making sure that promotional materials for CCBs depict women as well as men, through to more complex and long term efforts to change social conditions to allow women to participate on an equal footing with men.

Local issues, local solutions A function of this baseline report is to highlight the status of provision of public services in 2002 and the inequality of their delivery geographically and according to inequalities in society. It highlighted many deficiencies in public services in 2002, from the point of view of the public they are intended to service. But there were some encouraging signs. It was encouraging that despite lack of information at the time of the survey about CCBs, many ordinary citizens expressed interest in joining a CCB once they had an idea what CCBs were for. This impetus has also been reflected in the increasing rate of registration of CCBs throughout the country since the time of this data collection, even before the rules for CCB registration were promulgated. Willingness to join a CCB was nearly the only area in which ‘very vulnerable’ households did not lag behind less vulnerable households. If this willingness can be translated into actual active participation in CCBs it will indeed be a big step forward. It is clear from the findings that the issues about delivery of public services and participation in local democratic processes vary from place to place. This implies that as well as federal and provincial polices to support improvements in public services delivery, local solutions will need to be found. The development of local solutions can be facilitated by discussion of the findings from the baseline survey, disaggregated to district level, between elected representatives, government officials, service providers, and

“CCBs are like an ongoing welfare and a blessing.” Male focus group, Haripur

72

representatives of civil society. Already the process of seeking for solutions has begun by the feedback of some key findings from the district to separate male and female focus groups in each of the survey sites. As well as giving context to the quantitative findings, these groups started to suggest possible solutions to the problems highlighted. In some cases, citizens felt hesitant about how much difference they could make. But in other cases they were willing to get involved in monitoring public services and helping to solve particular local problems, and this bodes well for the emerging CCBs. Every community is different, yet each aspires to the same: a system that will hear their voice and address their needs.